Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2010-02-11

00:00:16 Quit matsl (Quit: Riece/3.1.2 XEmacs/21.5-b29 (garbanzo, linux))
00:00:39*gevaerts likes the idea of "Sure, you can have five fonts on your WPS, but you'll have fewer images then!" :)
00:01:14amiconnGiven the same total buffer size, a unified buffer makes better use of the available space
00:02:27 Quit Torne (Ping timeout: 256 seconds)
00:02:30 Join Torne [0] (torne@rockbox/developer/Torne)
00:03:20amiconnHmm, very interesting...
00:03:31amiconnThe higher levels are seeing less speedup
00:03:32***Saving seen data "./dancer.seen"
00:03:47*amiconn wonders what's going on there
00:06:29amiconnHmm, D2 uses no IRAM?
00:06:37*amiconn suspects caching issues then
00:07:06amiconnHow large is the TCC7801 cache?
00:08:15 Quit Galois (Ping timeout: 260 seconds)
00:08:20 Join piroko [0] (
00:09:02pirokoIt looks like the wiki section for ipod battery runtime hasn't been updated in a while. Anyone out there have a current approximate runtime for the 4G greyscale ipod?
00:09:15piroko(I guess I could just see how long mine lasts) :)
00:10:57gevaertsamiconn: from what I can see, 16K instruction and 16K data
00:11:55gevaertspiroko: that sounds like the best way. 2006 is indeed long ago...
00:15:37amiconnHmm, that should at least be large enough to not slow down -c4000
00:16:05 Quit bertrik (Quit: De groeten)
00:16:14amiconnAnyway, it's an overall speedup (and -c5000 isn't really important on that cpu)
00:19:20 Join Kitr88 [0] (
00:20:24 Join toffe82 [0] (
00:20:42 Join checker [0] (
00:21:16checkeris there a file size minimum on .sbs?
00:21:24 Quit Kitar|st (Ping timeout: 264 seconds)
00:23:21CIA-88New commit by 03amiconn (r24590): APE: Fused vector math for the filters on ARMv5te. Speedup on Cowon D2 is ~4% for -c2000..-c4000 (less for -c5000). Thanks to Frank Gevaerts for ...
00:23:40amiconngevaerts: Do you want to update the wiki, or shall I?
00:23:51gevaertsgo ahead
00:24:23checkerdo "|" characters mess up the .sbs for some reason?
00:24:46 Quit Frampis (Read error: Operation timed out)
00:24:51JdGordon_checker: yes
00:25:01JdGordon_use %| if you want a | to be displayed
00:25:20JdGordon_that should be on the wiki page?
00:26:02checker...... yes
00:26:06checkerthere it is :D
00:27:00 Quit Guest83017 (Read error: Operation timed out)
00:28:51 Join feisar [0] (
00:29:23 Nick feisar is now known as Guest38641 (
00:29:27 Quit Sajber^ (Quit: Leaving.)
00:30:01 Join Sajber^ [0] (
00:30:44 Join Frampis [0] (
00:31:14checkerwould a progress bar on the .sbs make file browsing much slower?
00:31:43JdGordon_it shuoldnt
00:31:56checkerscrolling text seems to take a toll
00:32:27 Quit Guest38641 (Read error: Operation timed out)
00:32:55JdGordon_scrolling text updates more frequently
00:39:20 Quit bmbl (Quit: Bye!)
00:39:24 Quit Frampis (Ping timeout: 264 seconds)
00:45:41 Join feisar_ [0] (
00:47:01 Quit jgarvey (Quit: Leaving)
00:49:24 Join Frampis [0] (
00:52:09 Join togetic [0] (~togetic@unaffiliated/ibuffy)
00:54:34 Quit soap (Remote host closed the connection)
01:01:15 Join soap [0] (~soap@rockbox/staff/soap)
01:04:41 Join Strife1989 [0] (
01:07:25 Quit Strife89 (Ping timeout: 240 seconds)
01:07:31 Nick fxb__ is now known as fxb (
01:13:35 Quit avacore (Ping timeout: 260 seconds)
01:13:37*amiconn is puzzled
01:14:37amiconnWhy does -c2000 (+19%) and -c3000 (+28%) become significantly faster when using fused vector math in the sim, but significantly slower for -c4000 (-13%) and -c5000 (-24%) ?
01:19:21 Join Casainho [0] (~chatzilla@
01:21:02 Nick Strife1989 is now known as Strife89 (
01:30:17 Join FlynDice [0] (
01:39:15 Quit thegeek_ (Read error: Connection reset by peer)
01:41:52 Join akur [0] (
01:42:30 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
01:42:40 Quit LambdaCalculus37 (Ping timeout: 245 seconds)
01:43:59 Join kramer3d_ [0] (~kramer@unaffiliated/kramer3d)
01:44:40 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
01:44:47 Quit linuxstb (Read error: Operation timed out)
01:46:10 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
01:48:40 Quit petur (Quit: Zzzz)
01:49:33 Quit BHSPitMonkey (Ping timeout: 256 seconds)
01:49:33 Quit Casainho (Ping timeout: 256 seconds)
01:49:47 Nick kramer3d_ is now known as kramer3d (~kramer@unaffiliated/kramer3d)
01:50:35 Join Galois [0] (
01:54:11 Join antil33t [0] (
02:02:20 Quit BlakeJohnson86 (Quit: Leaving.)
02:03:05Adnyxokramer mod?
02:03:34***Saving seen data "./dancer.seen"
02:03:47 Join BlakeJohnson86 [0] (~bjohnson@2002:1876:a27b:0:227:13ff:fe65:1262)
02:05:10 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
02:16:21 Quit LambdaCalculus37 (Quit: Fwump)
02:19:04 Quit kaniini (Ping timeout: 240 seconds)
02:25:25 Nick fxb is now known as fxb__ (
02:26:00 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
02:29:25 Quit Sajber^ (Read error: Connection reset by peer)
02:33:35 Quit linuxstb (Read error: Operation timed out)
02:37:58 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
02:44:42 Quit komputes (Quit: I haven't slept for ten days, because that would be too long.)
02:57:43 Quit Rob2222 (Quit: Rob2222)
03:01:25 Join drf|away [0] (
03:01:51drf|awaySo guys. I managed to get a v1 Sansa Fuze. However, I installed the bootloader, it said it installed correctly, and it's still booting the OF.
03:02:09 Quit akur (Quit: Leaving.)
03:04:12drf|awayoh never mind, I selected the wrong drive... I'm stupid
03:04:48 Join iq [0] (~chatzilla@unaffiliated/iq)
03:08:18 Part froggyman
03:12:36 Part drf|away
03:14:22 Quit kramer3d (Ping timeout: 252 seconds)
03:15:24 Quit Kitr88 (Read error: Connection reset by peer)
03:15:55 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
03:16:41 Quit moos (Remote host closed the connection)
03:18:51 Quit Adnyxo (Ping timeout: 265 seconds)
03:22:11 Join Rob2222 [0] (
03:27:01 Quit linuxstb (Ping timeout: 276 seconds)
03:32:42 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
03:40:52 Join docgnome [0] (~docgnome@
03:42:11 Join anewuser [0] (anewuser@unaffiliated/anewuser)
03:44:12docgnomesomeone recommend an cheap music player that can run rockbox?
03:51:00krazykit`basically every supported sandisk player
03:52:08docgnomek cool
03:52:29 Quit DerPapst (Quit: Leaving.)
03:53:26 Join CaptainKewl [0] (
03:54:19 Quit linuxstb (Ping timeout: 245 seconds)
03:56:26 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
03:59:11checker#progress bar %V|45|302|150|1|0|003031|C6CFD6|-| %pb|rockitlrspb.bmp|44|301|152|7| is there a problem with this? or i guess, what is the problem lol
03:59:23checkerfor .sbs
03:59:30 Part docgnome
04:02:56 Join dude187 [0] (
04:03:05checkerwait i see the image is now outside of the viewport
04:03:37***Saving seen data "./dancer.seen"
04:07:03checker#progress bar %V|44|301|150|7|0|003031|C6CFD6|-| %pb|rockitlrspb.bmp|45|300|150|5| ok this now...
04:07:07checkerstill doesnt work
04:11:17 Quit panni_ (Read error: Connection reset by peer)
04:12:29checkerx and y values are measured from within the viewport... duh. so 1,1 centers it and fixes it. solved.
04:14:04 Quit iq (Quit: ChatZilla 0.9.86 [Firefox 3.5.5/20091102152451])
04:20:02 Quit linuxstb (Read error: Operation timed out)
04:20:02 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
04:21:42 Quit TheSeven (Disconnected by services)
04:21:42 Join The_Seven [0] (~theseven@rockbox/developer/TheSeven)
04:21:42 Nick The_Seven is now known as TheSeven (~theseven@rockbox/developer/TheSeven)
04:22:34 Join earHurts [0] (
04:23:04earHurtsThe ipod video sumulator: how do I "reboot" it
04:23:36 Nick dmb is now known as d (~Dmb@unaffiliated/dmb)
04:24:04 Nick d is now known as ______dmb (~Dmb@unaffiliated/dmb)
04:28:12 Quit antil33t (Read error: Connection reset by peer)
04:30:53 Quit linuxstb (Ping timeout: 258 seconds)
04:31:40 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
04:33:46 Join S_a_i_n_t [0] (S_a_i_n_t@
04:34:10S_a_i_n_tJdGordon: ping
04:34:33 Join antil33t [0] (
04:34:38 Quit dude187 (Ping timeout: 276 seconds)
04:35:24 Join dude187 [0] (
04:37:29S_a_i_n_tr24581, Nano 1g, context menyu option "set as backdrop" fails to set the backdrop with the message "Backdrop Failed" the a known issue by now?
04:39:14S_a_i_n_talso, r24581, Nano1g, plugin "mandlebrot" hangs on a solid black screen and causes the player to need a hard-restart.
04:41:45 Quit antil33t (Read error: Connection reset by peer)
04:44:06JdGordonS_a_i_n_t: burp
04:44:53S_a_i_n_tdid you know about the backdrop failing to set via the context menu already?
04:45:26S_a_i_n_twell, now you do.
04:45:34JdGordonfile it
04:45:42S_a_i_n_tfails with message "backdrop failed"
04:45:50S_a_i_n_twill do.
04:45:51 Quit Barahir (Ping timeout: 265 seconds)
04:46:49checkeranyone know anything about a section of album art disappearing after key movement on the .wps on the gigabeat f20?
04:47:30 Join Barahir [0] (
04:47:56 Join antil33t [0] (
04:53:11S_a_i_n_tchecker: whick theme are you using?
04:53:51checkerone in the making :)
04:54:20S_a_i_n_tare you sure that it's not an overlapping viewport then?
04:54:29S_a_i_n_tseems quite possible to me
04:54:33earHurtsin the simulator, is there an easy way to print to stdout/stderr?
04:55:04checkerthey appear in random spots, its full screen width and about 15px high
04:55:25checkerrandom spots in the album art area*
04:55:30 Join webguest98 [0] (
04:56:21checkeri dont see any over lapping
04:56:21 Quit webguest98 (Client Quit)
04:57:03checkerit only happens which the volume indicator switchs to the next part of the strip bump, and when i fast forward or rewind
04:57:06S_a_i_n_tchecker: can you veriffy that it happens with Cabbiev2?
04:57:13S_a_i_n_t*the default WPS
04:57:16checkerlet me see
04:57:23S_a_i_n_tif not, I'd be looking at your code
04:57:44checkernope its good
04:58:20checkerit goes away after switching to the menu and back
04:58:27checkeron mine*
04:59:20S_a_i_n_twell, I'd be wanting to look at your WPS code then, as I suspect you've not done something correctly.
04:59:33S_a_i_n_tI could have a look at it, but can't promise anything.
04:59:42checkerk let me upload it somewhere
04:59:45S_a_i_n_tpastebin the WPS code
04:59:48 Join dtube [0] (
05:00:48*fleebailey33 points at desowin
05:00:51*fleebailey33 points at dtube
05:01:54checkerk done, but how do i link that to you?
05:02:03S_a_i_n_tBackdrop issue filed as FS #10996
05:02:50S_a_i_n_tk, I'll have a look patient tho :P
05:03:05 Quit anewuser (Ping timeout: 240 seconds)
05:09:28 Quit linuxstb (Ping timeout: 246 seconds)
05:10:42S_a_i_n_tchecker: Hmmmm, without knowing your image dimensions...I can't tell if there are any overlaps, the rest of the code *seems* sound (at a first glance)
05:11:21checkerthere shouldn't be any overlaps, i dont think thats the problem
05:11:57checkerbut its got to be theme related because cabbie doesnt do it...
05:12:00S_a_i_n_thave you run it in a SIM, or just on the device?
05:12:12checkerjust device
05:12:36 Quit MethoS- (Remote host closed the connection)
05:12:39S_a_i_n_tAha....I'll try the SIM myself, and see if that gives me any output about what's wrong.
05:13:08checkeryou want the images?
05:13:42S_a_i_n_tplease...upload the whole WPS ( is good)
05:14:29checkeryou want it in a zip
05:14:49S_a_i_n_tzip, 7z...whatever's good for you
05:15:07checkerok, 1sec
05:18:32checkerhope i got it all,
05:19:40S_a_i_n_tsweet, thanks...I'm pottering between jobs, so just stay online and I'll PM you shortly(-ish)
05:19:51checkeri g2g in 5min
05:20:06S_a_i_n_tPM me your email
05:20:12checkerhow do i pm
05:23:11 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
05:23:30 Quit n17ikh (Ping timeout: 245 seconds)
05:24:17 Quit checker (Quit: CGI:IRC (EOF))
05:28:22 Join n17ikh [0] (
05:33:29 Quit kramer3d (Ping timeout: 240 seconds)
05:35:09 Quit earHurts (Ping timeout: 265 seconds)
05:38:53 Quit Strife89 (Quit: Bed.)
05:40:04 Quit Horscht (Quit: Verlassend)
05:42:04 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
05:49:29 Quit vegtoruci (Ping timeout: 240 seconds)
06:01:11 Join kaniini [0] (
06:03:40***Saving seen data "./dancer.seen"
06:08:11 Quit kramer3d (Ping timeout: 252 seconds)
06:10:03 Quit Xerion (Read error: Connection reset by peer)
06:10:48 Join Xerion [0] (
06:22:31 Quit tha (Ping timeout: 265 seconds)
06:23:04 Join tha [0] (
06:30:18 Join framo4 [0] (
06:30:24 Quit FlynDice (Remote host closed the connection)
06:38:11 Quit framo4 (Quit: muuh)
06:38:23 Join FlynDice [0] (
06:39:21 Quit FlynDice (Remote host closed the connection)
06:39:25 Quit nimak (Read error: Connection reset by peer)
06:41:08 Join FlynDice [0] (
06:41:23 Quit linuxstb (Ping timeout: 256 seconds)
06:42:09 Join nima [0] (
06:57:29 Quit FlynDice (Remote host closed the connection)
06:58:48 Join FlynDice [0] (
06:59:20 Quit FlynDice (Remote host closed the connection)
07:04:17 Quit desowin (Ping timeout: 240 seconds)
07:04:25 Join FlynDice [0] (
07:04:56 Join desowin [0] (~desowin@atheme/member/desowin)
07:13:22 Join midgey [0] (~tjross@rockbox/developer/midgey)
07:15:26midgeysaratoga: (for the logs) ipod 4g test_codec results for mdctexp branch ->
07:16:36 Quit togetic (Quit: WeeChat 0.3.0)
07:17:16 Join togetic [0] (~togetic@unaffiliated/ibuffy)
07:20:19 Quit midgey (Quit: midgey)
07:20:28 Quit antil33t (Read error: Connection reset by peer)
07:22:11 Quit BHSPitMonkey (Ping timeout: 252 seconds)
07:25:18pixelmaS_a_i_n_t: mandelbrot has been renamed to fractals, if you still have mandelbrot on your player (and you mean that) it will be outdated and you should remove it. I'm not sure why it should show a black screen but maybe there was a bug in the version you installed last
07:26:18 Join antil33t [0] (
07:27:42 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
07:29:31S_a_i_n_tpixelma: I built it from current SVN r24581 I believe
07:29:41S_a_i_n_tthat's why I thought it was odd...
07:30:55S_a_i_n_tyes, 24581...that's how it compiled, *perhaps* something went wrong on my end, but I wanted to check first.
07:31:14pixelmayour new version does not contain a mandelbrot anymore and the one you have there is from before...
07:31:37pixelmabefore the rename I mean
07:34:03S_a_i_n_twhat I mean was, I'm pretty damn sure it compiled with both mandlebrot and fractals, and I deleted my .rockbox from my DAP before moving on the new *must* of messed up somewhere in the process of colpiling I guess. I think I'll do a complete new checkout.
07:35:46JdGordonanyone game enough to try multifont on target?
07:36:32S_a_i_n_tyes, I haven;t checked the Forums yet however...
07:36:46S_a_i_n_tI was unsure how the process went.
07:40:18 Join teru [0] ( after the font is declared to a slot, the viewport can then read as "%Vx|y|x|y|2|-|-|" "%Vx|y|x|y|3|-|-|" etc. up to 5?
07:40:57S_a_i_n_tIf so, I'll quickly hack up one of my existing WPS's
07:41:18S_a_i_n_tcool, I won't get cocky..I'll just try 2 user for now
07:41:26JdGordon1 is always th UI font for the screen, 2-X is custom fonts
07:41:35JdGordonX being unknown just yet
07:42:34JdGordonthe font must obviously be loaded before the %V line'll take some time to compile etc. but you'll know in 30mins or so.
07:43:11JdGordoncomment out firmware/powermanagements.c line 747 or it wont compile
07:43:24S_a_i_n_tsweet, thanks.
07:43:38JdGordonand dont use the version i uploaded earier, use the one from yesterday
07:43:42JdGordonthats one is broken :p the *first* post of the patch?
07:45:20S_a_i_n_tor second rather?
07:45:40S_a_i_n_tWednesday, 10 February 2010, 09:35 GMT+1
07:45:46S_a_i_n_tthat one?
07:45:47JdGordonneither :) the one I'm about to upload
07:45:55S_a_i_n_tah.....right :D
07:48:08JdGordongo for it
07:48:45S_a_i_n_t#ifdef HAVE_LCD_BITMAP
07:48:45S_a_i_n_t#endif ?
07:49:02JdGordondont touch anything. just compile it
07:49:02S_a_i_n_tooops, /* glyph_cache_save();*/
07:49:17S_a_i_n_tah...ok, :P
07:49:20 Quit FlynDice (Remote host closed the connection)
07:50:06S_a_i_n_tso...don;t comment out that line in firmware/powermanagement.c?
07:50:18S_a_i_n_tthat's what I was trying to confirm.
07:51:48S_a_i_n_t....sorry, do or don't edit 747 of firmware/powermanagement.c?
07:52:48 Join Zagor [0] (
07:52:48 Quit Zagor (Changing host)
07:52:48 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
07:53:09JdGordondont touch it
07:53:18JdGordonserisouly... it will rape your cat if you do!
07:56:20 Join FlynDice [0] (
07:56:34S_a_i_n_twell, compiling now....I'll let you know, or screendump a WPS for you whem I've finished
07:56:42S_a_i_n_tbut it's CygWin remember :P
07:59:40JdGordonthat patch will give you 7 fonts to play with or 6 if you have a target with a remote lcd
07:59:43 Part toffe82
08:01:00S_a_i_n_tHmmmmm, well, I'll try for two...just to be safe, and to stop me from writing a new WPS (even if it is a basic one :P ), I can just add 2 lines to my existing one, and change 2 numbers
08:01:45S_a_i_n_tactually, I can do 4 at once
08:02:02S_a_i_n_tso, bah...we'll se anyway. (fingers crossed)
08:02:18pixelmaS_a_i_n_t: could you check file dates of your mandelbrot and fractals.rock?
08:03:08 Join einhirn [0] (
08:03:41***Saving seen data "./dancer.seen"
08:04:57JdGordonS_a_i_n_t: let me know if you find it doing more disk access then usual
08:05:27JdGordonpixelma: any chance you can test the fm patch today? if not then I'll just go ahead and commit it and fix any bugs which get found
08:05:28 Quit nima (Read error: Connection reset by peer)
08:06:13 Join nima [0] (~nima@
08:06:19S_a_i_n_t11/02/2010 4:11am
08:06:39 Join stoffel [0] (
08:06:52S_a_i_n_tfor both, I doubt it's a code fault...I'm guessing something messed up while compiling.
08:07:32S_a_i_n_tI've checked out a new/clean source, if it doesn;t happen again...I'll put it down to a dirty SVN
08:22:01JdGordonUnhelpful: ping?
08:22:50S_a_i_n_tJdGordon: after specifying "%Fl|2|12-fontname|" then the following viewport can use that font, but if a viewport is declared after that which just specifies "1" -user, I wouldn't have to use the %Fl line to load that right? I assume o and 1 are always loaded?
08:23:18S_a_i_n_t*0 and 1
08:23:39JdGordon0 is always sysfont, 1 is always the ui font (for that screen), 2-9 are the available skin fonts
08:24:45S_a_i_n_tsweet, and any viewport that isn't 0 or 1 needs the %Fl line? or they all do now?
08:26:12JdGordonthe %Fl is to load a font, its not linked to a viewport
08:26:24JdGordonif you want a use a font other than 0,1 you need a Fl
08:26:26JdGordonotherwise you dont
08:26:48S_a_i_n_tright, I get it now...they all load at once, not per viewport.
08:27:14JdGordonyes, and before you ask, each skin needs those lines, and the numbers can be different
08:27:26JdGordoneach font will only be loaded once, even if both screens want it
08:28:12 Join stooo [0] (
08:32:42 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
08:35:15 Quit stooo (Quit: Leaving.)
08:49:28 Join petur [0] (~petur@rockbox/developer/petur)
08:52:58pixelmaI find this way of setting more fonts very complicated
08:53:32pixelmaJdGordon: I could in the evening if you remind me
08:54:26JdGordonwhich way of setting fonts?
08:54:46 Quit amiconn (Disconnected by services)
08:54:49 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
08:55:10 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
08:55:26 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
08:55:26 Quit pixelma (Disconnected by services)
08:55:35 Quit BHSPitMonkey (*.net *.split)
08:55:36 Quit togetic (*.net *.split)
08:55:36 Quit n17ikh (*.net *.split)
08:55:36 Quit Topy44 (*.net *.split)
08:55:36 Quit mc2739 (*.net *.split)
08:55:36 Quit Zambezi (*.net *.split)
08:55:36 Quit sinuc (*.net *.split)
08:55:36 Quit crashd (*.net *.split)
08:55:36 Quit xavieran (*.net *.split)
08:55:36 Quit jvd (*.net *.split)
08:55:36 Quit Kopfgeldjaeger (*.net *.split)
08:55:41 Join Zambezi [0] (Zulu@
08:55:44 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
08:57:11 Quit linuxstb (Ping timeout: 265 seconds)
08:57:41 Join TopyMobile [0] (
08:58:52 Join crashd [0] (
08:58:58 Join sinuc [0] (
08:59:19 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
08:59:30 Join jvd [0] (~syscrash@poipu/developer/syscrash)
09:01:07 Join flydutch [0] (
09:02:44 Join n17ikh [0] (
09:04:03 Join togetic [0] (~togetic@unaffiliated/ibuffy)
09:04:56 Join xavieran [0] (
09:05:01JdGordonpixelma: whats a better way to set the fonts
09:06:03*S_a_i_n_t thinks it's fine....uploading "proof of concept" now. 5 fonts onscreen :D
09:07:23 Join ender` [0] (
09:08:17S_a_i_n_tJdGordon: Anyone:
09:08:25S_a_i_n_t5 user fonts onscreen.
09:08:38*JdGordon sees garbage text
09:08:52JdGordonor was that just scrolling?
09:09:33 Join BHSPitMonkey [0] (
09:09:34 Join Topy44 [0] (
09:09:37*S_a_i_n_t just noticed that *was* scrolling while I screendumped
09:09:45S_a_i_n_tweird it's always that line though
09:10:31 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
09:10:34S_a_i_n_tI'll upload a better pic, it was just because it's scrolling
09:10:36pixelmaJdGordon: maybe with a font browser or a "set as" context menu, not sure. My main problem with this is that it looks like it could be quite complicated if you want to use e.g. Clip themes as a rwps for the Iriver remote and vice versa which should be possible with multifont
09:11:01JdGordonpixelma: wouldnt that make it harder
09:11:21S_a_i_n_tpixelma: it's not complicated at all...just set all the fonts you want to use right at the start of the WPS...that's what I did.
09:11:24S_a_i_n_tworks fine
09:11:25JdGordonthis way the fonts are directly linked to the sckin
09:12:23pixelmawell, can you have two "font #1"s?
09:12:41pixelmaor #2 etc.
09:12:57S_a_i_n_tworks fine for me, I can't imagine a better system...a context menu (esp for the WPS) would be chaos...and not sure how you'd do it either.
09:13:05JdGordonfont 3 in the wps and font 3 in the sbs are not necessarily the same font
09:13:19JdGordonthis way there is no guessing. You use the font number at the top of this skin file
09:14:02JdGordonthe numbers in the skins are just identifiers. I could change it to letters to make it more obvious, but thats a bit complicated
09:14:12S_a_i_n_t <- *BETTER* proof of concept.... JdGordon, start gloating now :-D
09:16:00JdGordonmaybe the font id's should be letters like image id's. Just to make it hugely obvious what they are
09:16:05pixelmaso for a target with an lcd remote you could use - I don't know - 10 different fonts? (assuming 5 different ones for each screen)?
09:16:19pixelmaat the same time that would be
09:16:36JdGordonfont 1 will always be the ui font from the setting for that screen though
09:16:49S_a_i_n_tJdGordon: I think numbers are more then it fits with the current 0 and 1 font scheme
09:17:15S_a_i_n_tit's easy to remember that "next user font is 2, then 3 and so on" etc.
09:17:41JdGordonnot easier than a,b,c
09:17:54JdGordonor s,m,l
09:18:14pixelmaJdGordon: ok, then there's "only" one problem - you can't set a user font for the remote independently yet
09:18:22S_a_i_n_tbut it *does* fit with the current scheme of the fonts
09:18:53JdGordonpixelma: yes, and I have no idea how to do that. I'm tihnking a context menu option on .fnt files, but thats not really nice
09:23:40pixelmaevery Clip theme would still have problems with viewports selecting the user font. I would expect a remote user font setting on targets with a remote (much like other remote lcd settings). And then an rwps would use that
09:24:46S_a_i_n_tthis is still early days...
09:25:22JdGordonpixelma: if you manually set the remote font using the "remote font" cfg setting, then that will just work
09:25:33pixelmaas long as it isn't set explicitly it could be the same as the main target user font
09:26:02 Quit CaptainKewl (Quit: ( :: NoNameScript 4.22 :: ))
09:26:04pixelmathere is no "remote font" yet
09:26:23JdGordonbut even then. if the skin works on the clip it will work on the h300 remote
09:26:27JdGordonthere is in the patch
09:27:13pixelmacfg only settings :\
09:27:40 Quit liar (Ping timeout: 245 seconds)
09:27:48JdGordonpixelma: yes, and I have no idea how to do that. I'm tihnking a context menu option on .fnt files, but thats not really nice
09:28:42JdGordonthe font browser right now uses the file browser to load the .fnt directly into the setting. I need to somehow make it smarter
09:29:05pixelmaI was thinking a "Remote font" browser
09:29:35JdGordonyes, but thats not possible right now
09:29:37pixelmayou have a "Remote backlight timeout" setting too
09:30:10JdGordonwe can easily use the file browser for the skins because the main and remote skins have different extensions
09:30:21 Join avacore [0] (
09:32:28pixelmaif you introduced the remote font setting, what is the problem making a second font browser for it? Really just out of curiosity
09:34:42 Quit stoffel (Remote host closed the connection)
09:34:53JdGordonthe equivilant of windows opening .xls in excell happens. .fnt is hardcoded to always be loaded into the main screen font setting. I need to break that assosiation somehow, and keep it simple
09:35:20JdGordonit's not a matter of not wanting the browser, its a matter of not being able to use it
09:38:08pixelmaand how does your cfg remote font setting work? I'm sorry, I can't imagine this now
09:39:09JdGordonits identical to the font setting, just with the word remote on the front
09:39:34JdGordonremote font: /.rockbox/fonts/16-GNU-Unifont.fnt
09:39:43pixelmaI meant if this works for .fnt why would a browser not work?
09:40:45JdGordonI'm not sure how else to explain it
09:41:11JdGordonthe font browser is literally just opening up the file browser in the font directory
09:41:32JdGordonpressing select tells the browser to do the select action on the filetype, which is .fnt
09:48:42rasherThere's some context for the file browser
09:48:53rasherIt could know that it's browsing for remote fonts
09:49:10rasherand not just call the open function
09:50:42 Join liar [0] (
09:52:33*JdGordon needs to think about it for a bit
09:52:46 Quit Zambezi (Changing host)
09:52:47 Join Zambezi [0] (Zulu@unaffiliated/zambezi)
09:59:28 Quit dtube (Quit: Leaving)
10:01:40 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
10:03:44***Saving seen data "./dancer.seen"
10:12:08 Join _zic [0] (
10:12:33 Quit AndyI (Ping timeout: 248 seconds)
10:13:36 Quit linuxstb (Ping timeout: 264 seconds)
10:17:46 Join AndyI [0] (~pasha_int@
10:30:02 Join FOAD_ [0] (
10:31:21 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
10:33:03 Quit linuxstb (Read error: Operation timed out)
10:33:30 Quit FOAD (Ping timeout: 245 seconds)
10:33:30 Nick FOAD_ is now known as FOAD (
10:39:06 Quit BHSPitMonkey (Remote host closed the connection)
10:49:00 Quit sinuc (Changing host)
10:49:01 Join sinuc [0] (~sinuc@unaffiliated/sinuc)
11:01:30 Join ranmachan [0] (
11:03:11amiconnJdGordon: You could make the .fnt association show a mini-menu on multiscreen targets
11:03:13ranmachanHi there. Any sansa c200v2 experts in here?
11:03:41ranmachanI followed the instructions on for "Installation for Unusable (other AMS Sansas)"
11:05:01ranmachanThe dualboot would still always load the OF, so I removed the 'boot of if USB is plugged in' check and recompiled mkamsboot
11:05:38ranmachanNow it tries to boot rockbox, but apparently panics, blinking out an error code with the two leds.
11:05:47ranmachanIf I poweron while holding right it boots the OF
11:06:01ranmachanAre those panic codes documented somewhere?
11:06:29ranmachanI assume it's bootloade-c200v2.sansa that is panicing here?
11:22:20 Join Casainho [0] (~chatzilla@
11:23:57 Join FOAD__ [0] (
11:27:36 Quit FOAD (Ping timeout: 256 seconds)
11:27:36 Nick FOAD__ is now known as FOAD (
11:41:08CIA-88New commit by 03teru (r24591): fix text scrolling handling in do_menu, set_time_screen and time_screen. ...
11:47:14*_zic really thinks that the usb stack should be enabled by default in the next stable release, as it works just *fine*
11:47:25_zicwhy is it still not in 3.5 ?
11:47:41topikwhere does it work *fine* ?
11:47:42gevaertsit is, unless you say which target you're talking about
11:48:02_zicipod nano 1st gen
11:48:18Torne_zic: because the ipods do not charge properly in Rockbox
11:48:31Torneand it's possible for them to end up *discharging* if you use them to transfer files over USB for an extended period
11:49:25Tornealso it's generally believed to be measurably slower than the original firmware
11:49:33Tornethough this might not actually be true for all models :)
11:49:42_zicyes, that used to be something like several weeks ago, but yesterday my device charged well while connected to the computer
11:49:54Torne_zic: i can assure you it *doens't* charge properly
11:50:10Torneat best it will charge at something like one eighth the speed of the original firmware
11:50:25Torneit depends what you are doing and what your settings are
11:50:49Tornethe hard disk ipods can easily discharge at quite a rapid rate while plugged in if you are copying files; that's probably not likely on the flash-based ones but they still charge very slowly
11:52:37Torneif you leave it on charge overnight you won't notice :)
11:52:53Tornebut in the original firmware you can generally recharge any of the ipods in well under two hours, or even under one
11:52:56Tornefrom flat to full
11:53:14Torne(well, ok, probably from 10-90)
11:53:41gevaertsTorne: while the rockbox USB stack is indeed measurably slower than the OF on some ipods, I claim that it's only 10% to 20% slower. The disk drivers however...
11:53:46*gevaerts hides
11:53:59Tornegevaerts: you mean the part that's blame-able on USB? :)
11:54:06Tornerather than the actual end user observable peformance? :)
11:54:11Tornethat's very helpful. *g*
11:54:19gevaertsTorne: well yes :)
11:54:29Tornei tried to make the disk faster!
11:54:35Tornebut it doesn't bloody work :)
11:54:44gevaertsWe should do some new measurements. I don't know the full impact of your changes
11:54:54Torneyes, we probably should
11:55:01Tornethe WRITE MULTIPLE change might've helped
11:55:06Torneshould've helped :)
11:55:32TorneDMA definately made read performance much better over USB, possibly it's now comparable to the OF even
11:55:43Tornebut people are generally more concerned about USB write speed on DAPs
11:56:11Torneif you can suggest a good benchmarking methodology i will experiment :)
11:56:36gevaertsI usually just use dd
11:57:56gevaerts"dd if=/dev/sdipod of=/tmp/ipod.img bs=64k count=100 && dd if=/tmp/ipod.img of=/dev/sdipod bs=64k count=100"
11:58:20gevaertsAlways do a quick read before writing to spin the disk
11:58:45*Torne notes, incidentally, that the disk performance isn't quite linear :)
11:59:03gevaertsI know, I don't care :)
11:59:10Tornethe beginning is the fastest bit
11:59:20Tornei guess that's probably what we do want to compare, though
11:59:35Tornethe slowdown over the rest of the surface is uninteresting because that's a fact of mechanics
12:00:00Torneit limits all disk access impartially :)
12:00:29gevaertsIt might be useful to measure at the end of the disk to see what happens if speed isn't USB-bound anymore
12:00:54Tornei have a test_disk patch that measures read speed across the entire surface (sampling at regular intervals)
12:01:18Tornei haven't actually put it in because it's ugly and relies on exporting storage_read_sectors to pluginapi
12:01:30Tornealso it has an option to do the same for writes which destroys your data :)
12:02:03Torne.oO( is there a valid case for adding storage_(read|write)_sectors to pluginapi? )
12:03:48***Saving seen data "./dancer.seen"
12:04:02gevaertsI'd say yes
12:04:13gevaertsA partitioning and formatting plugin has its uses
12:04:23 Join Sajber^ [0] (
12:04:34gevaertspossibly even an fsck plugin
12:04:43gevaertsmostly on archoses though
12:05:09 Nick fxb__ is now known as fxb (
12:05:44gevaertsThe standard answer of "it's a lot faster from your PC" isn't valid if you can't access the full disk over USB
12:05:45 Join watto [0] (~watto@
12:11:17Tornei might tidy it up at some point then
12:13:25 Join FOAD__ [0] (
12:16:59 Quit FOAD (Ping timeout: 252 seconds)
12:16:59 Nick FOAD__ is now known as FOAD (
12:18:58 Quit JdGordon_ (Ping timeout: 248 seconds)
12:19:34 Quit perfectdrug_ (Quit: perfectdrug_)
12:32:29 Join MethoS- [0] (~clemens@
12:34:36ZagorC64 mp3 decoder:
12:36:17ranmachanFWIW I've reported this on the rockbox sansa subforum now:
12:42:39 Part watto
12:51:38 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
12:52:27 Join matsl [0] (
12:52:52 Nick fxb is now known as fxb__ (
12:54:00 Join Kitar|st [0] (
13:05:02 Join kugel [0] (~kugel@rockbox/developer/kugel)
13:05:40kugelranmachan: don't we explicitely say the c200v2 is *unusable* and unsupported?
13:07:27kugelthat means we know that there are problems and don't want them reported on the tracker (and not on the forum anyway)
13:09:47ranmachankugel: Well there are other reports on the forum and the official devel thread is also there:
13:10:10ranmachanI can repost it there if you think that's more appropriate
13:10:13kugelthe development thread is different
13:13:35ranmachanHmm, looks like you or someone deleted, so I'll make a new post in the devel thread if I may
13:13:38 Join watto [0] (~watto@
13:14:30 Join Adnyxo [0] (
13:16:02 Nick YPSY is now known as Ypsy (
13:50:08S_a_i_n_twhen someone tals about the process "syncing a patch against current svn"...what exactly does this involve? If it's relatively simple, I'd really like to get FS #5111 - Ipod piezo driver to apply cleanly again.
13:52:48kugelS_a_i_n_t: open the patch file, go to the hunk(s) that failed (each file to be patched has hunks which start with the @@ x,y a,b @@ line), open the original file and compare what changed
13:53:52S_a_i_n_tthanks....*seems* relatively simple at theory anyway.
13:54:27kugelalternatively, you can checkout a revision where you know it works, apply, and then svn up to latest. then svn handles syncing but that may fail (but it's usually easier to resolve)
13:54:29S_a_i_n_tI'm, debating whether or not I need to get some sleep first to tackle this with minimal frustration however :-p
13:55:23S_a_i_n_tkugle...I actually wondered if it would cheaply work that way...thanks for that actually. I'll try that method first I believe :P
13:57:58S_a_i_n_tIf I revert my current tree to a previous revision, will it revert my previously applied patches as well? Or will it try to be smart about those also?
13:58:18S_a_i_n_tDo I need to revert my patches first is what I'm asking I guess.
13:58:55gevaertsS_a_i_n_t: short answer to all your questions: "it depends"
14:00:22S_a_i_n_tah....right. One of playing it safe and reverting patches beforehand could potentially avoid some grief?
14:00:32S_a_i_n_tI mean, I can always re-apply them.
14:00:45gevaertsif they still apply properly, yes :)
14:00:57gevaertsWhich gets you back to the "syncing a patch" question
14:01:06S_a_i_n_twell...there is that. :P
14:01:33S_a_i_n_tI'm sure I'll cross that bridge when I get to it.
14:02:31S_a_i_n_tI only need to revert back to 241XX-ish....<sarcasm>surely not much has changed since then </sarcasm>
14:03:49***Saving seen data "./dancer.seen"
14:05:09CIA-88New commit by 03gevaerts (r24592): Bump plugin API version. This should have been done in r24587. Also, because the new function is added halfway. the minimum api has to be bumped too/ ...
14:08:34 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
14:10:52 Part watto
14:12:35 Join Casainho_ [0] (
14:15:29 Quit Casainho (Ping timeout: 252 seconds)
14:15:32 Nick Casainho_ is now known as Casainho (
14:18:07 Quit MethoS- (Ping timeout: 256 seconds)
14:18:49 Join Psirus [0] (
14:18:49 Part Psirus
14:22:45S_a_i_n_tok...first snag I've hit. I'm getting "Conflict discovered in 'apps/lang/english.lang'. Select: (p) postpone, (df) diff-full, (e) edit, )mc) mine-conflict, (tc) theirs conflict, (s) show all options.
14:22:53S_a_i_n_twhat should I do?
14:23:18 Join TopyMobile_ [0] (
14:26:35dionoeaand try to fix it
14:26:36CIA-88New commit by 03mc2739 (r24593): Fix FS #10997 - Clipv1 Quickscreen - left button cancels instead of toggling shuffle
14:26:41 Quit TopyMobile (Ping timeout: 240 seconds)
14:27:15 Quit Casainho (Ping timeout: 245 seconds)
14:33:06 Quit Zarggg (Read error: Connection reset by peer)
14:33:23 Join Zarggg [0] (~zarggg@2001:0:4137:9e74:0:4388:beb1:ba3d)
14:34:13S_a_i_n_tHmmmm...I believe I may of found an easier option...choosing (df) and then (r) seemed to go smoothly for all all just depends on if it compiles or not.
14:34:44dionoeawhat does the r bit do ?
14:35:05 Join kugel_ [0] (
14:35:15 Quit kugel (Disconnected by services)
14:35:21 Nick kugel_ is now known as kugel (
14:35:25 Quit kugel (Changing host)
14:35:25 Join kugel [0] (~kugel@rockbox/developer/kugel)
14:35:30S_a_i_n_tatte,pts to merge
14:35:51 Quit mikroflops (Ping timeout: 260 seconds)
14:36:17 Join archivator [0] (
14:39:07CIA-88New commit by 03mc2739 (r24594): Correct Clip manual quickscreen keymap
14:43:14 Join efyx_ [0] (
14:46:39 Join MethoS- [0] (~clemens@
14:52:00S_a_i_n_tHmmmm....well, that didn't work.
14:52:26 Join DerPapst [0] (
14:52:26Zagorbefore anyone asks, yes there appears to be some network issues to the server
14:52:26 Join Will___ [0] (
14:52:26pixelmamaybe add it to the topic?
14:52:26Will___is the site down?
14:52:26Topic"Yes, there are network issues | 3.5 has been released! | Please read before speaking: | Please direct offtopic/social chat to #rockbox-community | This channel is logged at" by ChanServ (ChanServ@services.)
14:52:26pixelmaZagor just mentioned some network issues
14:52:26 Quit logbot (Ping timeout: 265 seconds)
14:52:26***ERROR: (Closing Link: (Ping timeout: 265 seconds)) from
14:52:26***Saving seen data "./dancer.seen"
14:52:28***Started Dancer V4.16
14:52:28***Connected to on port 6667
14:52:28***Logfile for #rockbox started
14:52:29Mode"logbot :+i" by logbot
14:52:30CtcpVersion from frigg!~frigg@freenode/utility-bot/frigg
14:52:31***Server message 501: 'logbot :Unknown MODE flag'
14:52:31 Join logbot [0] (
14:52:31 Join Will___ [0] (
14:52:31 Join DerPapst [0] (
14:52:31 Join MethoS- [0] (~clemens@
14:52:31 Join efyx_ [0] (
14:52:31 Join archivator [0] (
14:52:31 Join kugel [0] (~kugel@rockbox/developer/kugel)
14:52:31 Join Zarggg [0] (~zarggg@2001:0:4137:9e74:0:4388:beb1:ba3d)
14:52:31 Join TopyMobile_ [0] (
14:52:31 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
14:52:31 Join Adnyxo [0] (
14:52:31 Join Kitar|st [0] (
14:52:31 Join matsl [0] (
14:52:31 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
14:52:31 Join FOAD [0] (
14:52:31 Join Sajber^ [0] (
14:52:31 Join ranmachan [0] (
14:52:31 Join AndyI [0] (~pasha_int@
14:52:31 Join _zic [0] (
14:52:31 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
14:52:31 Join liar [0] (
14:52:31 Join avacore [0] (
14:52:31 Join Topy44 [0] (
14:52:31 Join ender` [0] (
14:52:31 Join xavieran [0] (
14:52:31 Join togetic [0] (~togetic@unaffiliated/ibuffy)
14:52:31 Join n17ikh [0] (
14:52:31 Join flydutch [0] (
14:52:31 Join jvd [0] (~syscrash@poipu/developer/syscrash)
14:52:31 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
14:52:31 Join sinuc [0] (~sinuc@unaffiliated/sinuc)
14:52:31 Join crashd [0] (
14:52:31 Join Zambezi [0] (Zulu@unaffiliated/zambezi)
14:52:31 Join pixelma [0] (quassel@rockbox/staff/pixelma)
14:52:31 Join amiconn [0] (quassel@rockbox/developer/amiconn)
14:52:31 Join petur [0] (~petur@rockbox/developer/petur)
14:52:31 Join nima [0] (~nima@
14:52:31 Join einhirn [0] (
14:52:31 Join FlynDice [0] (
14:52:31 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
14:52:31 Join teru [0] (
14:52:31 Join antil33t [0] (
14:52:31 Join desowin [0] (~desowin@atheme/member/desowin)
14:52:31 Join tha [0] (
14:52:31 Join Xerion [0] (
14:52:31 Join kaniini [0] (
14:52:31 Join Barahir [0] (
14:52:31 Join dude187 [0] (
14:52:31 Join S_a_i_n_t [0] (S_a_i_n_t@
14:52:31 Join TheSeven [0] (~theseven@rockbox/developer/TheSeven)
14:52:31 Join Rob2222 [0] (
14:52:31 Join BlakeJohnson86 [0] (~bjohnson@2002:1876:a27b:0:227:13ff:fe65:1262)
14:52:31 Join Galois [0] (
14:52:31 Join soap [0] (~soap@rockbox/staff/soap)
14:52:31 Join Frampis [0] (
14:52:31 Join feisar_ [0] (
14:52:31 Join piroko [0] (
14:52:31 Join Torne [0] (torne@rockbox/developer/Torne)
14:52:31 Join fleebailey33 [0] (fleebailey@unaffiliated/fleebailey33)
14:52:31 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
14:52:31 Join HBK [0] (
14:52:31 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-uxeftoofekqjpbkk)
14:52:31 Join Tomis [0] (~Tomis@
14:52:31 Join Tuplis [0] (
14:52:31 Join shaggy-h [0] (
14:52:31 Join Hadaka [0] (
14:52:31 Join evilnick [0] (~evilnick@rockbox/staff/evilnick)
14:52:31 Join dmb_ [0] (~Dmb@unaffiliated/dmb)
14:52:31 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
14:52:31 Join ______dmb [0] (~Dmb@unaffiliated/dmb)
14:52:31 Join gevaerts [0] (~fg@rockbox/developer/gevaerts)
14:52:31 Join jd [0] (~jd@Wikipedia/HellDragon)
14:52:31 Join flyback [0] (
14:52:31 Join maraz [0] (
14:52:31 Join krazykit` [0] (
14:52:31 Join zumbi [0] (~zumbi@
14:52:31 Join jordan` [0] (~jordan@
14:52:31 Join killan [0] (
14:52:31 Join parafin [0] (
14:52:31 Join jfc [0] (
14:52:31 Join tmzt [0] (
14:52:31 Join martian67 [0] (~martian67@about/linux/regular/martian67)
14:52:31 Join Res1 [0] (
14:52:31 Join ThomasAH [0] (
14:52:31 Join @Llorean [0] (~DarkkOne@rockbox/user/Llorean)
14:52:31 Join bzed [0] (
14:52:31 Join linuxguy3 [0] (
14:52:31 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
14:52:31 Join elcan [0] (
14:52:31 Join rasher [0] (~rasher@rockbox/developer/rasher)
14:52:31 Join pippijn [0] (~pippijn@unixcorps/staff/pippijn)
14:52:31 Join lyngaas [0] (
14:52:31 Join lostlogic [0] (~lostlogic@rockbox/developer/lostlogic)
14:52:31 Join yosafbridge [0] (
14:52:31 Join CIA-88 [0] (~CIA@
14:52:31 Join jhulst [0] (
14:52:31 Join GodEater [0] (~bibble@rockbox/staff/GodEater)
14:52:31 Join beta2k [0] (
14:52:31 Join rvvs89 [0] (~ivo@pdpc/supporter/base/rvvs89)
14:52:31 Join Ypsy [0] (
14:52:31 Join koniiiik [0] (
14:52:31 Join AlexP [0] (~ap@rockbox/staff/AlexP)
14:52:31 Join advcomp2019_ [0] (~advcomp20@unaffiliated/advcomp2019)
14:52:31 Join Kohlrabi [0] (
14:52:31 Join topik [0] (awesome@
14:52:31 Join simabeis [0] (~simabeis@
14:52:31 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
14:52:31 Join aevin [0] (eivindsy@unaffiliated/aevin)
14:52:31 Join BeFalou [0] (~mamutoi@unaffiliated/befalou)
14:52:31 Join zu [0] (~zu@
14:52:31 Join alexbobp [0] (~alex@
14:52:31 Join ps-auxw [0] (~arneb@2001:470:c807:0:1532:4e5f:2ad3:4123)
14:52:31 Join blithe [0] (~blithe@
14:52:31 Join MagusG [0] (
14:52:31 Join rphillips [0] (
14:52:31 Join scorche [0] (~scorche@rockbox/administrator/scorche)
14:52:31 Join jobec [0] (
14:52:31 Join kadoban [0] (
14:52:31 Join chaos [0] (~chaos@gentoo/user/ch4os)
14:52:31 Join crwl [0] (
14:52:31 Join @ChanServ [0] (ChanServ@services.)
14:52:31 Join fxb__ [0] (
14:52:31 Join preglow [0] (
14:52:31 Join dionoea [0] (~dionoea@videolan/developer/dionoea)
14:52:31 Join pjm0616 [0] (~user@
14:52:31 Join ehntoo [0] (
14:52:31 Join scorche|sh [0] (
14:52:31 Join rb_voicebot [0] (
14:52:31 Join B4gder [0] (~daniel@rockbox/developer/bagder)
14:52:33Zagorback now
14:53:13 Join vegtoruci [0] (
14:56:06Will___(Sansa e200v1) Is it normal for RB to display "maintenance mode" when I connect USB?
14:58:00mc2739Will___: does it say "maintenance mode" or "Multimedia Mode"?
14:58:40Will___I'm asking because I just fell asleep with my player connected, and when I woke up (7 hours later), it is still powered-on. I thought RB powered-down when charging was done...?
14:59:18mc2739no, Rockbox will not turn off when connected to USB
15:00:18Will___can 7 hours of connection damage my player?
15:00:44 Join perfectdrug [0] (
15:00:54*S_a_i_n_t gives up (the attempt at...) syncong FS #5111 for now....
15:01:07S_a_i_n_t*syncing rather
15:03:10 Quit Will___ (Quit: ChatZilla 0.9.86 [Firefox 3.5.7/20091221164558])
15:04:31 Join dfkt [0] (dfkt@unaffiliated/dfkt)
15:05:35 Join dfkt_ [0] (dfkt@unaffiliated/dfkt)
15:06:17 Quit kramer3d (Quit: Leaving)
15:08:57 Quit matsl (Quit: Riece/3.1.2 XEmacs/21.5-b29 (garbanzo, linux))
15:09:19 Quit Sajber^ (Quit: Leaving.)
15:09:42 Join Sajber^ [0] (
15:11:32 Join watto [0] (~watto@
15:11:50 Quit dfkt_ (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
15:14:59kugelS_a_i_n_t: you can't just ignore conflicts
15:15:21S_a_i_n_tI realised
15:15:42kugeland even successful compilation doesn't imply correctness
15:15:46Torneresyncing a language file is basically the most simple and trivial thing you can do, though :)
15:15:55Tornefar easier than resyncing actual code
15:15:57kugelthat's true
15:16:13Tornetake the added language strings from the patch
15:16:19S_a_i_n_tI'm evidently far too tired for this right now...I'll have another look at it in the morning.
15:16:22Tornepaste them on the end of the latest lang file from svn
15:18:26S_a_i_n_tIIRC, there more conflicts than just the lang file...but there's a rather large % of chance that I read it incorrectly also.
15:18:33CIA-88New commit by 03teru (r24595): color_picker: a bit of rework for color_picker. ...
15:18:43S_a_i_n_ts/there/there were/
15:20:23Tornewell, that may be, and that would be harder, yes
15:20:30Tornebut you only mentioned english.lang so hey :)
15:20:37 Join Casainho [0] (
15:21:14perfectdrugwould someone mind committing this small thing:
15:21:22S_a_i_n_tthat was just the first one I hit...there were5 scattered around apps and firmware
15:21:23 Quit vegtoruci (Ping timeout: 276 seconds)
15:22:09 Join mikroflops [0] (
15:22:46 Join evilnick_B [0] (~0c140464@gateway/web/freenode/x-ifvdaffvyzhtgtua)
15:32:48*kugel ports his favorite theme to the fuze
15:34:02 Quit Tuplis (Read error: Operation timed out)
15:34:35rasherIs it DancePuffDuo?
15:35:01kugelno, this was the one I ported from the fuze to the e200 ;)
15:35:44kugellook at the e200 themes sorted by download count; the first one :p
15:36:08 Join Tuplis [0] (
15:36:37UnhelpfulJdGordon: what was it? i'm not going to be around for a long stretch until tonight, but i read logs... and quassel makes sure i know if somebody tries to talk to me. :P
15:39:14kugelperfectdrug: do you think your png is better than the existing one (except for the black screen)?
15:39:56kugelI don't care much afterall
15:40:19 Join funman [0] (~fun@rockbox/developer/funman)
15:40:24kugelfunman: hi :)
15:40:42perfectdrugyes i think so, it is just minor but when you created the svg you look differently at things :D
15:41:02funmanranmachan: ping
15:41:23 Quit Adnyxo (Ping timeout: 240 seconds)
15:41:58ranmachanfunman: Yes?
15:43:09ranmachanfunma: IIRC With the check in it would always boot into OF
15:43:28funmancan you try something on your c200v2 : add ascodec_write(AS3514_DCDC15,1); in the lcd_enable() function of lcd-c200_c200v2.c like it is done in lcd-ssd1303.c for the clip
15:44:38perfectdrugbluebroth3r, domonky: is it possible for rbutil to check filesystems for e.g. the irivers too? i talked to someone who used it h140 as external drive and formated it NTFS, rbutil install went fine but the firmwareupgrade does obviously not work, it may be nice if the util told him that there is something wrong.
15:44:39ranmachanLet's see...
15:45:09kugelperfectdrug: that's for www/ right?
15:45:23perfectdrugi think so yes
15:45:24funmanlcd not turning on is a different issue from dualboot, thanks for finding what was wrong!
15:45:50ranmachanYou're welcome :)
15:46:19ranmachanI just looked at dualboot and thougt that looks like the most likely place to cause it to always boot into OF
15:46:34kugelperfectdrug: I don't have a checkout right now, I'll try to do it later if I don't forget it. I want to finish my theme first
15:47:14perfectdrugkugel: ok thanks:) no rush
15:47:20ranmachanfunman: Should I inserted it before the lcd_send_command(R_STANDBY_OFF... ?
15:47:33funmanranmachan: it's there as a safety in case the button check doesn't work for some reason
15:47:58funmanranmachan: before
15:48:13ranmachanOk, I was assuming it was rather for convenience since the USB stuff is not yet working :)
15:48:48kugelfunman: after my exams I'll try to commit some fuzev2 code
15:49:14funmani'll try to get a Clip+ sometimes soon to replace my bricked Clipv2
15:49:33ranmachanBTW, do you know if the 7 test pads on provide JTAG?
15:49:51dionoeafunman: did you keep track of the number of clips you bricked ?
15:49:56ranmachanI was wondering about that before, but never got around to test it.
15:50:18kugelfunman: I have another fuzev2 if you want it
15:50:19funmanranmachan: nope i think it was only tried on m200v4 and e200v2 (you can find info in the forum thread, perhaps around pages 30)
15:50:26funmandionoea: 4
15:50:34funmanbut only 3 while rockboxing
15:51:16ranmachanUnfortunately while I'd have the time to test right now, the equipment is about 10000Km away :)
15:51:23funmankugel: if you can write lcd code we'll be at the same point on all targets so i don't need one
15:51:47 Join piotrekm [0] (
15:51:56 Quit piotrekm (Changing host)
15:51:56 Join piotrekm [0] (~pm@unaffiliated/piotrekm)
15:52:12kugelthe lcd code is mostly the same as on fuzev1
15:52:32kugelI've disassembled it, but haven't written code from it yet
15:52:53ranmachanfunman: Should be enough to rebuild the bootloader, right?
15:53:03funmanranmachan: i remember my c200v2 lcd didn't work sometimes in the bootloader, but always in the loaded binary, but backlight always worked
15:53:20funmanranmachan: according to your post it seems to load the binary just fine so i would rebuild only rockbox.sansa
15:53:24dionoeafunman: and did you get free replacements?
15:53:34funmandionoea: twice
15:53:55funmani didn't try to replace the clips i had opened
15:54:30dionoeajust use some glue and some paint before sending them back :D
15:58:21kugelI've successfully sent an opened fuze back :)
15:59:26ranmachanfunman: No, that doesn't seem to help.
16:00:02funmanif you use ascodec_write(...,15) ?
16:02:52CIA-88New commit by 03mc2739 (r24596): Updated Russian Translation ...
16:02:53funmanyou said the button light is on ?
16:03:31ranmachanNo change.
16:04:05ranmachanYes, first only the center button (second stage loader?), goes off, then center + menu (rockbox.sansa?)
16:04:22funmani don't remember there were 2 lights
16:05:29funmanthe backlight is off, or on but not showing any pixels ?
16:05:42ranmachanBacklight looks off, even in a totally dark room.
16:06:42funmanin the OF code i only see GPIOA_PIN5 being modified
16:06:44ranmachanranma/S6000696.AVI"> (13MB)
16:08:44funmandefinitely the backligtjh
16:09:49funmancould it be related to what bertrik and FlynDice were saying the other day, about current over a certain voltage
16:09:54ranmachanYeah, I just tried shining a strong flashlight on the LCD.
16:10:05ranmachanI can see the splash that way.
16:10:18funmancool :)
16:10:19ranmachanIt's still rebooting in a loop for some reason
16:10:31funmanhm that's another bug
16:10:33 Quit Casainho (Quit: ChatZilla 0.9.86 [Firefox 3.5.3/20091020122601])
16:10:42funmanif you plug it on usb, can you see the splash as well ?
16:11:14ranmachanSince I removed the USB check that should make a difference, right?
16:12:20ranmachanWhile only the center light is on I can see 4 Lines, the last one 'loading rockbox.sansa'
16:12:35ranmachanThen the splash and with revision and after that reboot
16:13:13 Join jgarvey [0] (
16:14:01ranmachanIt's the same when I power it on by plugin into USB
16:14:12funmanin ascodec-target.h : change (0<<6) in AS3514_CP_DCDC3_SETTING to (1<<6)
16:14:15ranmachan(Though the button blinking is slightly differnte then)
16:14:20 Quit teru (Quit: Quit)
16:14:55ranmachanWhich one? The one in arm/as3525 I presume?
16:15:51ranmachanLeave the write at ,15 or go back to ,1?
16:16:29funmanremove this one, i don't think it's needed at all
16:16:52funmanand hmm it was buggy, the other bits should be preserved
16:17:38funmanhm no it's not the same register, there are only 5 bits in DCDC15
16:19:47ranmachanNo change
16:20:56funmanthen i don't know, you should disassemble the c200v2 OF and look for what we miss
16:22:33ranmachanAny particular offset where I should start looking?
16:23:06 Quit S_a_i_n_t (Quit: [St.] has exited mIRC)
16:24:00funmanon version 03.02.05, the void backlight(int onoff) function i know is at 0x4128
16:26:00 Quit n17ikh (Ping timeout: 245 seconds)
16:28:50ranmachanIn the main firmware block according to
16:31:14 Join n17ikh [0] (
16:31:16funmanyou can use utils/AMS/hacking to extract it to "firmware" (be careful it will dump around 20 files in the current directory)
16:33:53 Quit funman (Quit: free(random());)
16:35:53 Quit Zambezi (Read error: Connection timed out)
16:44:04 Quit Zagor (Quit: Clint excited)
16:51:55 Join thegeek [0] (
16:52:30***Saving seen data "./dancer.seen"
16:56:10 Nick evilnick_B is now known as weevilnick (~0c140464@gateway/web/freenode/x-ifvdaffvyzhtgtua)
16:56:27 Join fyrestorm [0] (
17:10:54 Join pamaury [0] (
17:17:32 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
17:19:33 Join panni_ [0] (
17:23:19saratogaranmachan: if you are interested in the JTAG, it might be pretty useful to look at
17:30:40ranmachansaratoga: Right now I'm more interested in looking at the GPIOs from within rockbox :)
17:31:10ranmachanI managed to stop the rebootloop by disabling USB detection in the usb driver
17:31:24ranmachanLooks like the GPIOs might be different on my c200v2
17:32:00ranmachanI'd like to look at the GPIO state to see if something changes when I plug in usb...
17:33:04ranmachanmp3 playback: check
17:39:09 Join Zambezi [0] (Zulu@
17:40:05ranmachanHey, debug->view io ports looks promising :)
17:43:39ranmachanHmm, no change when plugging in USB though
17:43:49 Join toffe82 [0] (~chatzilla@
17:44:42kugelranmachan: you should try writing to that pin
17:45:03kugeli.e. if 1 would be inserted, try writing 0 to it before reading
17:49:03ranmachanFirst I'll try writing to A0,A1,A4 in addition to A5 and look if the backlight turns on :)
17:49:40bluebroth3rperfectdrug: should be possible.
17:55:01 Join anewuser [0] (anewuser@unaffiliated/anewuser)
17:58:27 Quit Zambezi (Changing host)
17:58:28 Join Zambezi [0] (Zulu@unaffiliated/zambezi)
18:00:03 Quit petur (Quit: work->home)
18:01:16 Nick weevilnick is now known as evilnick_B (~0c140464@gateway/web/freenode/x-ifvdaffvyzhtgtua)
18:01:28 Quit evilnick_B (Changing host)
18:01:28 Join evilnick_B [0] (~0c140464@rockbox/staff/evilnick)
18:04:31ranmachanSuddenly: Backlight! Now which GPIO is the question....
18:06:32kugelranmachan: maybe you need to only write to that gpio at init; for example the fuze and e200 have identical buttons basically, except that the fuze needs A1 set for the scrollwheel to work at all
18:06:40 Join komputes [0] (~komputes@ubuntu/member/komputes)
18:07:29 Quit GeekShadow (Read error: Connection reset by peer)
18:07:39ranmachankugel: Likely not, the backlight stayed on even when rockbox disabled the lcd (I only modified _on, not _off)
18:07:59 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
18:08:20 Join bzed_ [0] (
18:08:36kugelso does your c200v2 have a completely other pin for backlight or is A5 also needed
18:09:01ranmachanThat's what I'm trying to find out right now :)
18:10:51 Quit bzed_ (Remote host closed the connection)
18:10:52 Quit bzed (Ping timeout: 258 seconds)
18:10:54ranmachanLooks like it's controlled by A7, A5 is not needed.
18:12:12 Join bzed [0] (
18:16:17ranmachanInteresting: If I plug/unplug USB GPIOD changes from 1F to 9F for a short period
18:17:26 Join GeekShado_ [0] (
18:17:34ranmachanOk, that's just because buttonlight is turned on (menu light)
18:17:46kugelbertrik has a c200v2, you should talk to him so he can verify if you want it fixed in svn
18:17:57ranmachanHowever even though I've disabled the usb detect function it reacts to usb
18:18:38 Quit GeekShadow (Read error: Connection reset by peer)
18:22:58ranmachanFor now I've reported my findings on the devel forum thread:
18:23:11ranmachanAnd I'm going to sleep now, it's past 2am :)
18:27:09 Join moos [0] (moos@rockbox/staff/moos)
18:29:21pamaurydoes anyone here have a simulator build so he can try a 5sec manipulation ?
18:29:56pamauryIt's just a matter of taking a random file and displaying its properties. On my computer it crashes the sim
18:30:56JdGordonamiconn: yeah, a mini menu could maybe work. "set as main screen font" and "set as remote screen font:?
18:31:44JdGordonpixelma: reminder to please test the radio skin patch this evening
18:32:34TorneJdGordon: you mentioned hte multifont thing working for plugins, but not being sure if this was useful
18:32:35 Join moos_ [0] (
18:32:39Tornei have a use for multifont in plugins :)
18:32:46Tornefrotz requires text styles
18:33:00 Quit togetic (Quit: WeeChat 0.3.0)
18:33:17JdGordonI didnt say it wouldnt be useful. I just said I thught that plugins shouldnt require fonts
18:33:41 Join togetic [0] (~togetic@unaffiliated/ibuffy)
18:33:42Tornetbh frotz is going ot have quite picky requirements wrt. fonts
18:33:56Torneideally frotz wants to be able to do regular text, bold, italic, and monospace
18:34:02Tornewell, minimally, in fact
18:34:09Torneideally it wants any arbitrary combination of those :)
18:34:22kugelwho put the colon into the "Next Track:" lang string?
18:34:24Torneat the moment i'm intending to use the user font as regular text, and sysfont as monospace
18:34:35Tornebut tha tmight look weird for user fonts over a certain size :)
18:34:43 Quit moos (Ping timeout: 260 seconds)
18:34:44Torneand doesn't give bold/italics.
18:34:46 Nick moos_ is now known as moos (
18:35:00kugelpamaury: seems to work here
18:35:22kugelmy build is pretty old though
18:35:30kugelJdGordon: did you? :(
18:35:58pamaurykugel: strange, I'm at svn HEAD (or nearly), DEBUG, logf, e20 sim. what is your config ?
18:35:58TorneJdGordon: so i guess in an ideal world i'd want to be able to work out a set of fonts that could do some of that. dunno how plausible tha tis, though
18:36:08kugeltomers did
18:36:13Torneat the moment only the wps can load fonts, yes?
18:36:30kugelpamaury: no idea actually :)
18:37:01kugelno logf, fuze sim is what I know. I can't tell if I have patches applied
18:37:35 Quit zu (Ping timeout: 260 seconds)
18:37:36kugelI just run the sim for porting my theme without recompiling, i.e. it's the state where I left it off at last
18:38:31pamauryHum, it crashes in the properties plugin, on a strcmp, strange. Can someone else try ? (Torne ? JdGordon ?)
18:38:56JdGordonwhat am I trying?
18:39:21 Quit moos (Ping timeout: 256 seconds)
18:39:45pamauryIf you have a simulator build, could you try to display the properties of a file or a folder and see if it crashes, please
18:40:06 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
18:40:50 Join phanboy4 [0] (
18:40:52*kugel ponders making an sbs version
18:41:56 Join bertrik [0] (
18:42:10JdGordonno crash here
18:42:18pamauryHum, perhaps my build just got corrupted...
18:42:31JdGordonsvn + multifont patch
18:43:16kugeldomonoky: ping
18:43:44 Join moos [0] (moos@rockbox/staff/moos)
18:43:46kugelwhat the test environment for the theme site again?
18:44:18domonokykugel: i have a (hackish) vmware image to test theme site changes.
18:44:53domonokyi could upload the current image somewhere if you want.
18:45:18kugelI'm not sure I want it. I don't have vmware installed (and didn't plan to)
18:45:40gevaertskugel: virtualbox should be able to handle it
18:45:44JdGordonyou uploaded that to my server somewhere right? I can make it avilable
18:46:07pamauryOk, it works, something went wrong during the compilation it seems
18:46:13domonokyJdGordon: yes i did, but that copy is a bit outdated..">
18:46:25JdGordonbah :p 7z
18:47:04domonokythe problem with the vmware image is, that it isnt exactly the same as the theme server, so it needs some changes and hacks. (but is good enough for my own tests).
18:48:34kugelok, I would like to have a look
18:50:34domonokyoki, if you wait a bit, i will make a current image available.
18:51:32kugelI want to have a look at keeping the download count when updating
18:51:56kugelotherwise if I update my theme I lose the (overall) 1st place in the download count stats
18:52:32***Saving seen data "./dancer.seen"
18:53:19*JdGordon has a small epiphany
18:53:40JdGordonplugins can cause skin fonts to be unloaded and then reuse that buffer :p
18:54:14JdGordonso even if the skin uses all 7 user fonts, the plugins can use just as many. and if they both want the same then there is no special handling
18:55:04gevaertsso that would just cause some reloading on plugin entry/exit I guess?
18:55:38kugelcouldn't you write a wrapper for plugins so that they don't reuse existing buffers?
18:56:41domonokykugel: keeping download counters sounds good. ~40min till i uploaded the image.
18:57:13JdGordonkugel: yes, thats how it willl be done
18:57:35*JdGordon runs to work
19:02:31 Join zu [0] (
19:06:03kugelwhat the heck
19:06:08kugelI'm getting "Invalid dirname: .rockbox/wps/BlacknBlue_Glass/ (should be classic_statusbar.)" all over
19:06:40kugeloh nevermind
19:08:36 Quit bluebroth3r (Ping timeout: 265 seconds)
19:09:51 Join bluebrother [0] (
19:09:51 Quit bluebrother (Changing host)
19:09:51 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
19:13:11 Quit _zic (Quit: Ex-Chat)
19:15:12bluebrotherdomonoky: have you tried installing the bootloader on an ipod using windows lately?
19:15:30domonokybluebrother: i dont have a (working) ipod :-)
19:15:59bluebrothertoo bad. There seems to be a problem in rbutil (including the latest release :( )
19:17:19 Join Omlet [0] (
19:18:26domonokymaybe someone else here has a ipod and windows and can test it ?
19:19:16domonokybluebrother: did you upload the fixed macosx binary, or is the one on the download server still broken ?
19:20:00bluebrotherit's still broken.
19:20:36bluebrotherI'm on the ipod issue right now, as I assume that there are more windows users around
19:21:15 Join _zic1 [0] (
19:21:46 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
19:22:56bluebrotherdomonoky: I'm wondering if there's a way to prevent such issues in the future. Unfortunately testing is quite hard due to the device dependencies :(
19:24:27*Unhelpful pokes jdgordon with a stick
19:24:54domonokyautomatically testing the bootloader installs is quit impossible. but the mac issue should be preventable.
19:26:28domonokybut if we some day have our testing results page and include rbutil into it, we might find most issues before we release :-)
19:26:33gevaertsdomonoky: actually, with pamaury's usb-in-sim work, I think that automatically testing the bootloader installs isn't entirely impossible these days :)
19:27:14*bluebrother thinks he found the issue and has to blame himself :(
19:28:04domonokygevaerts: the sims dont use bootloaders, so we might be able to simulate what ipodpatcher/sansapatcher does, but we still dont know if the result works.
19:29:07pamaurygevaerts, domonoky: what do you want to test with the bootloader ?
19:29:12Unhelpfulsim... os == "booatloader"
19:29:13gevaertsdomonoky: true, but I'd assume that whoever releases the bootloader actually tests it, so the only thing you have to validate when testing the install is that the resulting disk image is exactly right
19:30:08*bluebrother still gets a segmentation fault on exit
19:30:50pamaurygevaerts: any plan to commit close-handles-on-plugin-exit one day ?
19:32:13pamauryHas anyone seen Slasheri recently ? How often does he go on the channel ?
19:33:16gevaertspamaury: hm, maybe we should just commit it right now
19:33:25bluebrotherpamaury: you can ask logbot
19:33:36pamaurybluebrother: how ?
19:33:47bluebrotherpamaury: /query logbot seen Slasheri
19:34:18bluebrothertry /query logbot help. Just did that myself to figure out how it worked ;-)
19:34:30*gevaerts used to not like the splash in that patch, but he thinks he changed his mind
19:35:09pamaurygevaerts: did I used a splash ? I used a yesno gui, no ?
19:35:30pamaurybluebrother: /query logbot help doesn't do anything for me :(
19:36:07 Quit toffe82 (Read error: Connection reset by peer)
19:36:27pamauryAh, ok.
19:37:30 Quit phanboy4 (Read error: Connection reset by peer)
19:37:32bluebrotherpamaury: sorry, use /msg
19:37:51pamauryI found out by myself ;)
19:39:28 Nick fxb__ is now known as fxb (
19:39:47*bluebrother reinstalls the bootloader for a felt 1000th time for tonight
19:40:44amiconnJdGordon: Yes, something like that. Only on multiscreen targets of course
19:41:29 Join Strife89 [0] (~michael@
19:41:44amiconnLoading a font already triggers a full redraw anyway, so that request shouldn't cause problems with screen contents
19:41:59 Join toffe82 [0] (~chatzilla@
19:43:10amiconnIt could even allow to cancel the .fnt loading as a 3rd option (although - single screen doesn't allow that)
19:46:03bluebrotherhow (un)stable is voice? I'm experiencing quite a lot data aborts since I started using the voice I generated on OS X
19:47:54 Quit piotrekm (Ping timeout: 256 seconds)
19:49:21 Join JdGordon_ [0] (~836b0052@gateway/web/freenode/x-netztotbusamhqac)
19:50:25 Quit dmb_ (Ping timeout: 246 seconds)
19:50:39domonokyvoice should be stable at least for older targets, but maybe someone made it unstable ?
19:50:52CIA-88New commit by 03bluebrother (r24597): Restore width of the Autodetect button.
19:51:30 Quit gevaerts (Disconnected by services)
19:51:39 Join gevaerts [0] (~fg@rockbox/developer/gevaerts)
19:53:53JdGordon_Unhelpful: hey, I wanted to ask if you know how safe the AA stuff is when playback is stopped?
19:53:58 Quit toffe82 (Read error: Connection reset by peer)
19:54:21JdGordon_the code looks like it shold be fine, but apparently it breaks on target (in the fm screen), so maybe there is something missing?
19:54:59 Quit zu (Ping timeout: 260 seconds)
19:59:43pixelmabluebrother: I have no problem with voice (I'm using my M5 the most currently).Just one freeze while browsing in a long time and the problem has been there for a looong time on coldfire (or at least the Iaudios) but isn't reproducible
20:00:46pixelmaJdGordon_: ok, going to test in a bit
20:01:09JdGordon_great :) thanks
20:01:35amiconnThe occasional freeze also happens on cf irivers
20:02:16gevaertspamaury: a yesno is a sort of splash in my view :)
20:02:36amiconn(and it's more annoying on them since resetting needs a paperclip)
20:03:14pamaurygevaerts: you're right but I really prefer the yesno in that case !
20:03:36DerPapsthey all. quick question... i have viewport and display a menu in it but i want to clear the backdrop before. any advice how to do that?
20:03:38 Join piotrekm [0] (
20:03:44 Join zu [0] (
20:03:53gevaertspamaury: anyway, do you want to commit, or shall I?
20:04:11JdGordon_DerPapst: why?
20:04:14gevaertsI think you spend more time on it than I did in total
20:04:45JdGordon_DerPapst: passing false to the last param of do_menu() will disable the theme and bckdrop
20:04:58pamauryOk. But just one thing: the patch enables the check by default and you said it should not be enabled for release builds. Do we put a memo somewhere for that ?
20:05:01DerPapstJdGordon_: because it looks real weird in the plugin :P
20:05:15DerPapstJdGordon_: thanks, i'll try
20:05:44JdGordon_are you using your own backdrop in the plugin? or a non full screen menu?
20:06:04gevaertspamaury: my view on that has also changed a bit :) We have three months before the next release, so by then there should be no more filehandle bugs, so leaving it enabled should be fine :)
20:06:29pamauryok :)
20:07:36pixelmabluebrother, pamaury: you can also query logbot and then "talk" to him. E.g. just type "help" into the query or "seen xyz"
20:07:56pamaurypixelma: that's what I did actually :)
20:08:07 Join Schmogel [0] (
20:08:55DerPapstJdGordon_: this is what the menu currently looks like the last parameter is false already btw
20:10:20JdGordon_DerPapst: woops, logic is backwards :p the last param shuold be true
20:10:34kugelyea, it's "hide_theme" so it should be true :)
20:10:37DerPapstheh... lets try. thanks
20:10:46JdGordon_are you using a viewport for the menu also?
20:11:21kugelJdGordon_: I wondered what's the matter with calling viewport_set_defaults on every theme_enable() call, is that needed?
20:11:59kugelI wouldn't expect this function to touch the viewport and all, and it also seems wasteful
20:12:36DerPapstyay, works. thank you :)
20:12:53JdGordon_kugel: which one?
20:12:53bluebrotherpixelma: I know :)
20:13:25kugelviewportmanager_theme_enable() calls viewport_set_defaults() for the passed viewport
20:13:49amiconnJdGordon: Btw, a mini-selector might be useful for other filetypes with more than one action as well
20:13:58bluebrotherpixelma: hmm, I'm on PP (mini2g). I'm somewhat wondering if the problem could be related to the voice −− otoh the voice file is just a voice file and I have no idea why a voice file should (even could) cause crashes
20:14:01DerPapstJdGordon_: yes. two viewports. one for the menu and one for the logo
20:14:14amiconn(bmp, ...)
20:14:32JdGordon_kugel: only if its passed in, the idea there is so you can setup the viewport+theme in one call instead of 2
20:15:07 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
20:15:07 Quit pixelma (Disconnected by services)
20:15:19 Quit amiconn (Disconnected by services)
20:15:20 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
20:15:23JdGordon_amiconn: well, how many extensions would this apply to? we don;t want to duplicate the open with code
20:15:24 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
20:15:39CIA-88New commit by 03pamaury (r24598): Commit FS #10889: detect file handles leaks in plugins and automatically close them on exit and warn the user.
20:15:42 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
20:16:15amiconnActually it may be possible to reuse the "open with" code. Just that the default would show less choices
20:16:44JdGordon_not in its current form
20:17:15 Quit flydutch (Quit: /* empty */)
20:17:21JdGordon_the code for it is heavily linked to the filetypes and only loads a plugin
20:17:59*amiconn also thinks that single-buffer font caching (possibly reusing the buflib) may pave the way to get lru caching going for other stuff
20:18:08amiconnE.g. voice on lowmem targets
20:21:41JdGordon_if we had malloc we would have far better memory usage overall!
20:22:01JdGordon_there are probably at least 5 differen buffer allocation implementations in the core already
20:22:10*JdGordon_ hides
20:22:26*gevaerts tracks JdGordon_ down so he can partially agree
20:22:49amiconnThat's not correct
20:22:57JdGordon_which part?
20:22:59 Nick Ypsy is now known as YPSY (
20:23:10amiconnmalloc needs a very different approach on memory usage.
20:23:13gevaertsThe big problem is that while I think that malloc might be OK, free() and realloc() aren't exactly straightforward
20:23:28amiconnBasically you must never use up all available memory.
20:23:56amiconnBut we do want that, i.e. use all available memory for buffering
20:24:28JdGordon_no, but we can allocate say 1MB for malloc and then the rest is audio
20:24:44amiconnThere are other problems with malloc, but this design issue is the most fundamental one
20:24:59amiconnSet aside 1MB of about 1.3MB free in total? Certainly not
20:25:15amiconnAnd that's archos - there are swcodec targets with even less free memory than that
20:25:58JdGordon_no, obviuosly i wasnt talking about low mem here
20:26:22amiconnFurther issues: memory fragmentation (you *cannot* avoid that on a system without mmu), increased complexity of all code needing buffers (you need to handle malloc failure) etc etc
20:28:22DerPapstthe reason why you have to reboot iPL after using some modules/programs...
20:28:31*amiconn really doesn't want his dap to display "malloc failure" in response to a user action
20:28:50amiconnIf I wanted that, I would have installed ipl on my ipods.... but no thanks
20:29:11pamaurymemory fragmentation in inherent to malloc, with or without mmu because you usually allocate small things.
20:29:13DerPapst*pffff* iPL is teh best!
20:31:11gevaertsWe could move to a handle-based model instead of the current pointer-based model for the bits that need lots of memory in a non-predictable way. That would fix the fragmentation issue. If we then also change buffering to handle changing amounts of available memory, we'd have most cases covered I think.
20:31:16gevaertsLots of work though
20:31:25 Join grndslm [0] (
20:31:49amiconngevaerts: That's why I mentioned buflib...
20:32:00 Join toffe82 [0] (~chatzilla@
20:32:20gevaertshm, yes. I wasn't paying attention...
20:32:55gevaertsAnyway, a pure traditional malloc()/free() model won't ever work properly
20:33:05*amiconn wonders whether it would be possible to extend buflib so that it handles *all* variable buffers
20:33:39amiconnI.e. font loading, voice, theming etc would all allocate from the same pool
20:33:47saratogawe can't really do malloc since we have no VM
20:33:58amiconnsaratoga: huh?
20:33:59saratogabut we could have a better dynamic memory allocation system
20:34:15saratogawithout VM you can't really implement free effectively
20:34:20kugelwe as well just use a real malloc then
20:34:20gevaertsYou still have to be careful. You don't want to allocate so much to voice+theme that you have to choose between a font and playback....
20:34:23amiconnYou *can* do malloc without virtual memory. It's just nasty
20:34:33amiconnAmigaOS is an example
20:34:52saratogasure you can malloc, you just can't expect free() to do what you want it to . . .
20:34:55kugelhow does tlsf malloc in the codeclib do it?
20:35:08gevaertsby luck I assume
20:35:10saratogakugel: it doesn't really do free correctly
20:35:23saratogait just depends on the codec being reinited every track
20:35:25amiconnfree() is possible, but you get fragmentation over time
20:35:38gevaertsIf you have lots of small allocations it will mostly work, but if you have some large ones, expect trouble
20:35:55 Nick fxb is now known as fxb__ (
20:35:57amiconnThat's why codecs can get away with it - they "just" have to survive one track
20:36:05saratogathe original malloc i tried for the codecs didn't work because it didn't get reinited on track change, which meant that eventually you ran out of buffer memory and it would crash
20:36:41saratogaif you played too many tracks of the same codec in a row (it was fine if you kept changing formats)
20:37:10amiconnAmigaOS does some clever tricks to minimize fragmentation (like doing small allocations from one end and large allocations from the other), but it can't be avoided completely
20:37:35saratogabut a system where you could request some (large) fixed size chunks might work, or one where you periodically have to give back the chunk so it can be defragmented
20:38:06saratogain a sense thats what we do now for various things (enabling tagcache or db), except we do it by forcing the user to reboot
20:38:37 Part zu
20:39:40JdGordon_buflib would still have problems with beign careful to not store pointers so they can be moved around right?
20:39:53 Join guesswho [0] (
20:40:08JdGordon_which means a MASSIVE amount of work for skins
20:40:10saratogayes, you have to update all pointers when you check back a block
20:40:13gevaertsJdGordon_: yes, but if you make sure that they get moved around, that won't be a problem for very long :)
20:40:39 Nick guesswho is now known as guesswhoooo (
20:40:43saratogaalternatively you can just request largish blocks (e.g. multiples of 64KB) and then try to minimize freeing
20:40:49amiconnsaratoga: The idea is to avoid pointers and use handles
20:40:52saratogathat should make fragmentation less of a problem
20:41:03saratogaamiconn: yeah thats what i figured
20:41:17saratogasounds like a great place to use OO
20:41:23saratogalets switch to c++
20:41:47gevaertssaratoga: uh? How does OO relate to this?
20:42:07saratogaput all this into objects and make them worry about the pointers
20:42:23guesswhooooumm.. evening.. is there a way to edit rockbox_default wps? I'd like to swap the places for artist and song name
20:42:26JdGordon_isnt that exactly what the handle does
20:42:38JdGordon_guesswhoooo: no, you can just copy that to a new .wps
20:42:46JdGordon_the code for that wps is in the wiki i tinhk
20:42:55guesswhooooah, I'll take a look from there
20:42:58saratogai assume you have to handle that yourself though, by dereferencing things like *handle+offset ?
20:43:19gevaertssaratoga: **handle I'd say
20:43:31JdGordon_void* get_handle_data(int handle)
20:44:18gevaertsThe old Mac way was to just have a pointer to an entry in a pointer table, which was located at the start of the heap
20:44:29 Join pamaury_ [0] (
20:45:23 Quit pamaury (Read error: Operation timed out)
20:45:27*amiconn thinks that buflib could in fact work quite well for the core, but it would need to be extended with a special buffer type
20:45:38 Quit krazykit` (Read error: Operation timed out)
20:45:42 Join krazykit [0] (
20:45:58saratogabuflib is buffering.c?
20:46:05amiconnA type that's marked as "can be thrown away when allocation isn't possible otherwise". This type would then be used by playback
20:46:37kugelsaratoga: nope, it's in the pluginlib
20:46:45kugelpictureflow uses it
20:46:49gevaertsyes and no. You don't want to throw away the bit you're decoding right now
20:47:08amiconnThis way playback wouldn't need to be stopped anymore e.g. when loading a larger theme
20:48:00gevaertsyes, and the dircache limit could probably go away
20:48:00kugelsounds like a shit lot of work (*cough* GSoC *cough*)
20:48:14amiconngevaerts: Yeah, actually those blocks need more differentiation
20:49:46gevaertsthe allocation code should ask the playback code I guess
20:49:53amiconnsaratoga: No, buflib is apps/plugins/lib/buflib.c
20:50:24amiconngevaerts: A callback? "Call me for futher information about this block"
20:50:33JdGordon_no more callbacks :(
20:50:50pixelmaJdGordon_: this build boots but I only get a blank radio screen (except statusbar) and if I'm lucky, I get some scrolling text from the menu... this is without having a fms set which I don't even know where to set it
20:50:57gevaertspossibly, yes
20:51:05amiconnNah, just a function in playback that buflib could call would be sufficient
20:51:22amiconnBlayback would then do the deallocation itself
20:51:37JdGordon_pixelma: ok, bugger
20:51:53CIA-88New commit by 03bluebrother (r24599): Restore the default message handler on application shutdown. Fixes sporadic segfaults on exit.
20:52:09JdGordon_would buflib allow us to use the entire buffer for everything? so there is no problem of choosing how much to allocate betwen it and playback?
20:52:10pixelmashould there be a menu option under theme settings?
20:52:15 Join jobec_ [0] (
20:52:18gevaertsActually, a function that asks playback to free some memory (with an argument to tell it how much is needed) should be enough. The allocator can then move things around to make the newly requested block fit
20:52:20JdGordon_no, fms: setting in the config
20:52:24amiconnJdGordon_: yes
20:52:32JdGordon_ok, sweet
20:52:36***Saving seen data "./dancer.seen"
20:52:39amiconnProbably except for code (plugins, codecs)
20:52:44pixelmanoooo :(
20:52:54JdGordon_its just not added yet
20:52:57 Join petur [0] (~petur@rockbox/developer/petur)
20:53:00gevaertsJdGordon_: yes, with (I guess) some rules so you have minima for some of them. You want to have more than 100 bytes left to load a WPS :)
20:53:12amiconnBut everything that is data would be completely flexible
20:53:42gevaertswell yes, as long as code is linked to a fixed address, you can't change that anyway
20:54:15JdGordon_well, if we had dynalically loadable plguins/codecs... :)
20:54:25*amiconn wonders how large a coff loader would be
20:54:35amiconnpreglow wanted to experiment with that iirc
20:55:19JdGordon_ok, 1 tiny possible issue. the skin allocator has 2 ways of allocing a block, one if the usualy "give me a block X size", the other is "give me the whole buffer and ill tell you how much I used when I'm done"
20:55:24 Join dmb [0] (~Dmb@unaffiliated/dmb)
20:55:28JdGordon_I assume buflib cant handle the second?
20:55:30 Quit jobec (Ping timeout: 246 seconds)
20:55:44amiconnIt would mean further core binsize increase, but at the same time it would do away with overlay plugins, and users not using plugins would waste zero buffer...
20:55:50gevaertsJdGordon_: maybe not right now (I haven't looked), but it should
20:56:04gevaertsThere's no real reason not to handle it
20:56:12guesswhoooohmm, can't seem to find the default wps
20:56:22JdGordon_(the second is used to alloc space for things of unkown sizes, like images (although that can be known) and the token array which *is* unkown)
20:56:22amiconnJdGordon_: The second case is rather bad
20:56:33JdGordon_yes, yes it is :p
20:57:12amiconnBasically it would either mean that skin loading might fail (if "give me all" is interpreted as "give me all that's free atm") or stopping playback
20:57:15 Join jobec [0] (
20:57:16gevaertsJdGordon_: don't expect too much performance, but realloc()-style things aren't impossible. It just may require lots of memmov()
20:57:21bluebrotherdomonoky: any thoughts on pushing a rbutil 1.2.5 shortly? I've found the line that crashes the bootloader installation
20:57:46domonokythen it would be good to make a new release :-)
20:57:59amiconngevaerts: buflib compacts anyway, using memmove
20:58:24JdGordon_well, those 2 cases can be removed if we read the bmp required size before actually loading it, and go to a linked list instead of a regular array for the tokens
20:58:27amiconnWhat might be tricky is to make the compactor not delay things too much. It might need to yield during compaction
20:58:28gevaertsJdGordon_: can that be changed to e.g. "give me 100K. I'll tell you to grow if I see I need more, and I'll shrink when done"?
20:58:47JdGordon_it can grow and shrink?
20:58:59bluebrotheranyone wants to update a rbutil translation?
20:59:45gevaertsJdGordon_: again, not *now* (I think...), but there's no reason why it can't be done
20:59:54 Quit jobec_ (Ping timeout: 252 seconds)
20:59:59JdGordon_ok, then it could work
21:00:40gevaertsYou check if there's enough free space after the buffer you want to grow, if there isn't, you shuffle buffers around until there is, and then you change the size field
21:01:08gevaertsShrinking is slightly easier, as it's guaranteed that there's room for it :)
21:01:55JdGordon_well, if there is any seriousness in wanting to do this, then I'll work the skin engine to use it as a proof of concept if someone does the buflib side of things
21:03:29gevaertsBasically the only differences with a full malloc() system are that (a) you use double indirection, so you have to be a bit more careful with pointers, and (b) allocation and resizing can be *a lot* slower. I'd never use this model for something that allocates and frees all the time, but for things like the skin buffer, dircache, or the database, it should work well
21:04:05amiconnPlus font buffer(s), voice, ...
21:04:11CIA-88New commit by 03b0hoon (r24600): Packard Bell Vibe 500: correct main keymaps, enable full keyboard editing, enable morse input.
21:04:19JdGordon_i.e all the allocation rockbox does :p
21:04:21pixelmaJdGordon_: blank screen with the fms: line and a file I had prepared (on target), can't tell for sure if the fms should work or not, I'd need to test in a sim but am not sure if it's worth it
21:04:30gevaerts"like" :). Yes, things that don't change without user intervention basically
21:04:45amiconnHmm, also the browser buffer
21:04:57JdGordon_pixelma: can you try loading a fms which just has some text and %wd?
21:04:59pixelmaor maybe prepare a super simple .fms
21:05:00amiconnNo more maunual limit setting
21:05:24amiconnThe browser would start with a small buffer and resize if a folder doesn't fit
21:05:40JdGordon_small being a reasoably size? or tiny?
21:05:47amiconn(of course not in single-entry steps...)
21:05:55kugelis this all threading safe? what if I get my data and another thread causes movement?
21:06:16amiconnkugel: You always have to re-evaluate handles after a yield()
21:06:24 Join FOAD_ [0] (
21:06:29amiconnThat's where cooperative threading helps a lot
21:06:49gevaertsyes, a handle based system with preemptive threading sounds like a nightmare
21:06:50 Join Tomis2 [0] (~Tomis@
21:06:52amiconnDual core might have special implications
21:07:12gevaertshm, there are some edge cases
21:07:13JdGordon_can we add a simple locking mecahnism for dual core?
21:07:36amiconnAs a first step, I'd just says the COP must never use handle allocated memory
21:07:37gevaertsDoes file I/O yield?
21:08:18gevaertsok, that needs some care then
21:08:31JdGordon_would a call to check if the pointer is still valid be a bad idea?
21:08:33 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
21:08:38amiconnIt yields e.g. when waiting for disk spinup
21:08:58gevaertsJdGordon_: how do you know that?
21:09:13amiconnJdGordon_: Why ask if it's still valid? That's slower than just re-requesting it
21:09:16JdGordon_pass the handle and pointer into the function
21:10:00 Quit FOAD (Ping timeout: 264 seconds)
21:10:00 Nick FOAD_ is now known as FOAD (
21:10:07gevaertsoh, you mean a pointer to the start of the buffer? Yes, that would work, but as amiconn says, why?
21:10:45amiconnI am imagining it as a function that replaces the dereferencing operator
21:11:00gevaertsI guess core functions that handle memory might need a way to lock a buffer
21:11:01 Quit Tomis (Ping timeout: 252 seconds)
21:11:02 Nick Tomis2 is now known as Tomis (~Tomis@
21:11:06amiconnI.e. instead of *pointer, you write geth(handle)
21:11:42amiconnIn C++ you could do this by operator overloading, I think (I never wrote C++ code myself...)
21:12:10gevaertsDepending of how you implement it, it can just be **handle, possibly with some volatile magic
21:12:12amiconnAh yes, buffer locking would be another option
21:12:42amiconnYou don't really need volatile in a cooperative threading environment, as long as no interrupts are involved
21:13:20CIA-88New commit by 03b0hoon (r24601): fix typo
21:13:27gevaertsbuffer locking only works if no buffer is more than 1/3 of the total buffer space I think (that number might be wrong, and I'm assuming that only one lock is active at a time. More locks need smaller maximum chinks)
21:13:43pixelmaJdGordon_: still a blank screen and the fms just containing: "%wd" "Station:" "%TnX" "%tf" (separate lines each)
21:13:59JdGordon_pixelma: change that X to 0
21:14:06amiconnHmm, of course it's just **handle
21:14:35JdGordon_oh yuck, deadlock issues if we add locking :(
21:15:18 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
21:15:51gevaertsI wouldn't add locking for general "users", only inside firmware/* functions that get passed a pointer to a buffer (are there more of those than just disk I/O?)
21:16:29 Join p3tur [0] (~petur@rockbox/developer/petur)
21:17:01JdGordon_well, if bufflib is in firmware/ we could just has a second set of i/o functions to use a handle instead?
21:17:12gevaertsyes, that's the other option
21:17:16 Quit petur (Disconnected by services)
21:17:20 Nick p3tur is now known as petur (~petur@rockbox/developer/petur)
21:17:23pixelmaJdGordon_: still blank - and the %wd doesn't seem to take effect. I can still see the statusbar or more precisely the
21:17:30JdGordon_int *handle, so the caller doesnt even need to allocate one first
21:17:55pixelmait was still set from my work on it back then
21:18:00gevaertsBuflib-allocated pointers would be forbidden for the other firmware/* functions then
21:18:39JdGordon_can anyone think of any other functions where thats a problem though?
21:19:33amiconnYou only need to consider functions which work on buflib-allocated memory
21:19:35JdGordon_pixelma: crud, ok, can you try the 2nd last patch instead of the latest one please?
21:19:59amiconnE.g. lcd update and rawing are no problem. The framebuffer is fixed size, no need to allocate that via buflib
21:20:12pixelmayeah, would be nicer if you could start numbering your patch versions btw.
21:20:16 Join tipi^ [0] (
21:20:31gevaertsYou also only need to consider functions that may yield
21:20:47amiconnYes, although that might be difficult to predict across targets
21:20:57JdGordon_lcd_set_backdrop() will be problematic
21:20:59amiconnE.g. lcd_update[_rect]() yields on some targets
21:21:31JdGordon_we can lock the buffer in the lcd_update()'s though
21:21:38amiconn(just an example for the difficulty - of course lcd_update() isn't a problem)
21:22:05amiconnJdGordon_: It would be all drawing functions, no the update
21:22:13JdGordon_it sort of is a problem, the yield could cause playback to buffer more data and move handles
21:22:34amiconnHence you need to re-evaluate handles after a yield...
21:23:04JdGordon_or add a yeild_and_lock() for formare so it doesnt cause problems 15 layers up
21:23:13amiconnThe drawing functions don't need locking
21:23:31gevaertsexactly. And then another thread can do file IO which also locks some bits, and then another thread does something, and as soon as you have lots of locks some thread asks for a new buffer (or a resize) that would require moving a locked buffer to be moved
21:23:33amiconnöcd_set_backdrop() needs to work differently though - it needs to store the handle
21:23:34JdGordon_the thing that calls it might need to reget its handles
21:23:59amiconnEach drawing function would then need to get the backdrop handle before starting to draw -> problem solved
21:24:08amiconnDrawing functions never yield half-way
21:24:10 Quit alexbobp (Ping timeout: 260 seconds)
21:24:15*gevaerts thought this was a good idea when we didn't have any details yet
21:25:18 Join p3tur [0] (~petur@rockbox/developer/petur)
21:25:20 Quit jd (Quit: Ω)
21:25:21 Quit petur (Disconnected by services)
21:25:25 Nick p3tur is now known as petur (~petur@rockbox/developer/petur)
21:25:26JdGordon_I'm talking about locking the whole buffer, not individual handles
21:25:39 Join jd [0] (
21:25:39 Quit jd (Changing host)
21:25:39 Join jd [0] (~jd@Wikipedia/HellDragon)
21:25:40JdGordon_nothing shuold block waiting for a new handle
21:25:47JdGordon_hence, no deadlocking
21:25:49 Join pyro_maniac_ [0] (
21:25:54 Part watto
21:26:17amiconnIf you lock the whole buffer and something wants to do an allocation, what should happen?
21:26:29JdGordon_that should never be allowed
21:26:30*amiconn thinks that locking is bad, and should be avoided altogether
21:26:35JdGordon_activily avoided
21:26:47JdGordon_or at least, we lock any movement
21:27:04amiconnThen you have to deal with failures...
21:27:26gevaertsno, you just have to make movement yield and wait
21:27:46amiconn-> deadlock
21:27:50JdGordon_not really. The bulk of allocations are done at init, worst case is file i/o waiting for a block
21:28:10 Quit anewuser (Quit: Another edition of chiptune gig WinterChip5! :O =ooo)
21:28:25gevaertswhy? Only firmware/* functions should ever lock, and only apps/* should ever allocate or change allocations
21:28:33JdGordon_^ bingo
21:28:51amiconnWell, firmware/ may also allocate
21:28:57JdGordon_a yield() in lcd_update() will never cause a forced alloc, buffering might ask to alloc, but it will be refused a handle
21:29:17JdGordon_i/o is different and wont do a full lock
21:29:40JdGordon_unless someone comes up with somewhere else not in the main thread which allocs, its not an issue
21:29:59amiconnPlayback isn't in the main thread
21:30:18amiconnDircache is yet another thread
21:30:23pixelmahmm... exchangint the firmware file should be enough on the Ondio to test the radio screen patch, or am I missing something?
21:30:26kugelJdGordon_: as a note, the multifont patch breaks checkwps so beware :p
21:30:27JdGordon_right, so it needs to make sure not to while(1) if the buffer is locked
21:30:31pixelmaexchanging too
21:30:42JdGordon_pixelma: yes, it should
21:30:50JdGordon_kugel: yeah, I know (or expected it anyway)
21:31:19*gevaerts somehow thinks that there won't be any deadlock, but he has no proof
21:31:37JdGordon_anything outside the main thread just does a non blocking handle request and yields if it fails
21:31:56JdGordon_those threads are not critical if the alloc waits longer (except maybe buffering)
21:32:07JdGordon_which would just lead to a playback stall, not a crash/lockup
21:32:07gevaertspossibly wrapped in a loop that sleeps until unlock
21:32:29JdGordon_no, it should be handled by the main thread loop properly
21:32:32gevaertsbuffering can't use the allocation anyway until the disk is ready for it
21:32:50gevaertsAnd non-disk locks should be *very* short
21:33:15JdGordon_as lnog as yield changes to a lock yield we shuold be safe
21:33:35CIA-88New commit by 03amiconn (r24602): Remove leftover from before make system rework.
21:33:36 Join fml [0] (
21:34:08amiconnlinuxstb: ^^ That's not the one for the standalone build, of course
21:34:21JdGordon_the COP is then the last major obstacle
21:34:59fmlJdGordon: hey! I still think UI fonts shouldnot get special treatment in the firmware. Just allocate the buffer at RB start and load the UI fonts (if specified) like all other skin fonts.
21:35:21 Quit yosafbridge (Quit: Coyote finally caught me)
21:35:22pixelmaJdGordon_: I get a blank screen with the previous version of the patch too
21:35:29JdGordon_ok thanks
21:35:34JdGordon_fml: I replied in the thread
21:35:41JdGordon_and I duisagree :)
21:35:48fmlJdGordon: we could even treat the sysfont the same way if it can be described as a "font", i.e. as a struct font
21:36:18JdGordon_it *is* a font struct
21:36:19 Join yosafbridge [0] (
21:36:20fmlJdGordon: yes, I saw your reply, but you agree with all I said actually.
21:37:14JdGordon_I said it wasnt worth the effort to move it, and it doesnt really make it cleaner
21:37:23fmlJdGordon: then we can handle the sysfont the same way. Just "load" it in the beginning so the ref gets 1. And then it will stay forever since noone will unload it.
21:37:27JdGordon_sysfont has to be handled differently, ui font may as well also be
21:38:09fmlJdGordon: why does it have? I don't think so. And it *will* make firmware cleaner.
21:38:36fmlWemy of course have different understanding of cleanness :-)
21:39:09JdGordon_no, because the setting MUST be loaded into the FONT_UI or FONT_SYS slots, we have no way to force that, and that font will change during a session
21:39:39JdGordon_and FONT_UIREMOTE for them
21:40:00 Quit yosafbridge (Client Quit)
21:40:04amiconnpamaury_: Do we really need this handle leaking check? It costs quite some binsize for what it does
21:40:08 Join yosafbridge [0] (
21:40:21amiconnImo the plugin author has to make sure his plugin doesn't leak handles
21:40:56 Join alexbobp [0] (~alex@
21:41:53pamaury_that was proposed by gevaerts :) I'm *just* the implementer.
21:41:53kugelhehe, I got a PM already for my plans to keep the download count when updating
21:42:12fmlJdGordon: we can. We can load them at the start so that they will get loaded at the right positions.
21:42:15amiconnIt could be useful for developers, but imo it should be disabled by default
21:42:32JdGordon_fml: and then how do you load a different font to that slot?
21:43:14 Quit fml (Quit: CGI:IRC (EOF))
21:43:39pamaury_amiconn: I also think it's useful for developpers but you're right, perhaps it should be disabled by default. Ask gevaerts, he thinks it should be enabled by default.
21:43:48 Join fml [0] (
21:44:10 Quit guesswhoooo (Quit: CGI:IRC)
21:44:51fmlJdGordon: we can have a replace function or something like that.
21:44:57amiconngevaerts: ^^
21:45:40JdGordon_fml: but then you are at the point where you are adding more code for that case than to just have the 4 line special case for them that it is now
21:46:07JdGordon_and then it even more so special cases those fonts
21:46:13fmlJdGordon: and 10K wasted memory for nothing
21:46:27JdGordon_remote font you talking about?
21:46:35pixelmainterresting. I never noticed that the radio kept playing in the USB screen on the OndioFM at least
21:46:40fml(besides other things)
21:47:01JdGordon_the remote buffer is the only thing wasted if its unused
21:47:10fmlpixelma: I think it's always been like that, on other targets too
21:47:34JdGordon_I was thinking about spitting the main screen remote buffer 50:10, but dont know a nice way to see if the split is needed
21:47:50JdGordon_without reloading both font files
21:47:54fmlJdGordon: and then we wouldn't have that special treatment for the NULL params which is not very nice IMO
21:48:27JdGordon_you just move the special casing out of firmware into more volatile code
21:48:31JdGordon_it doesnt make sense imo
21:48:34pixelmaI guess on the M5 the 5:1 split wouldb't be fair
21:49:13fmlJdGordon: btw: I've only looked at the patch, not the whole file. But I noticed that there are fd's (file descriptors) in the app level font descriptors (skin fonts) as well. Why that?
21:49:58fmlJdGordon: I won't be able to properly look at that until after the next week
21:50:16JdGordon_no, there is an id, not an fd
21:51:40kugeldomonoky: the guy was worried about people abusing the rating and dl count feature
21:51:52pixelmaalthough album art doesn't work (correctly) on the remote which reminds me that I should write some bug reports
21:52:16domonokykugel: he PMd me too..:-)
21:52:25fmlJdGordon: ah, right!
21:52:44kugelI'm not sure if dl count should be made unique
21:53:12kugelif I download the theme again after it was updated it should count again; but downloading every 2 seconds is surely too abusive
21:53:54domonokyits also not easy to prevent abuse of download counts and ratings. currently only the sum is store.
21:54:41kugelmaybe it could work like this: if an ip downloads/rates 3 times within 15min then lock it out for 2h (numbers could be changed)
21:56:03domonokyyou still need to store the ips in the db somewhere, which is probably not preferable. but maybe some cookie magic could help ? ( i dont know much about web things)
21:56:34 Quit Strife89 (Quit: Going home.)
21:56:58kugelme neither :(
21:57:08kugelbut cookes are easier to trick around
21:58:28JdGordon_do we really care that much if peopkle want to be stupid?
21:59:08domonokysure, but it might but the barrier high enough so we dont get much abuse, and it might be easier (if someone knows cookie magic)
21:59:40CIA-88New commit by 03kugel (r24603): Save download count when updating a theme
21:59:49 Join phanboy4 [0] (
22:01:12domonokykugel: commit looks good. i will make it live now :-)
22:01:23kugelJdGordon_: we not so much, but the themers do
22:01:29kugeldomonoky: thanks :)
22:01:55dionoeathe issue with ip restrictions is people behind NATs (or countries behind NATs)
22:02:35 Join Dhraakellian [0] (
22:03:09Dhraakellianwas USB support for the 5G iPod Video disabled in the 3.5 build?
22:03:24kugelthe captcha idea sounds bloody annoying if you ask me
22:04:20domonokycaptchas for rating are too much hassle i think. For reporting themes its fine.
22:04:35kugelthe guy suggested it for downloading too
22:04:44domonokyand we surely dont want captchas for downloads !
22:05:04dionoeacan't you just use cookies? (we use that for theme rating on the vlc skins download page and have never had any abuse, at least not significant enough to notice)
22:05:32domonokydionoea: that was my idea, but i dont know cookies.. maybe you could help out ? :-)
22:06:15kugelsure, it's just that cookies never convinced me for preventing abusing. and someone needs to code it of course
22:06:16pixelmaDhraakellian: I think it was disabled for all Ipods in the release because they still have issues with charging in Rockbox USB which can actually lead to discharge
22:06:28JdGordon_oh, just tell the themers to get stuffed, their ego's arnt that important that a few fake downloads will change anything
22:06:33dionoeadomonoky: in php its easy: $_COOKIE["somekey"] = somevalue;
22:06:53Dhraakellianpixelma: ah
22:07:15Dhraakellianwell, given that my Woot BoC iPod only gets ~2h battery life *anyway*
22:08:19JdGordon_AlexP: you wouldnt like to try the latest patch and let me know if you get a blank screen would you?
22:08:49pixelmahmm, I could test on my M5
22:09:10domonokydionoea: sounds easy :-)
22:09:32kugelsetcookie("prevent_abuse" ,"asd" ,time()+60*5 ); seems to do it too
22:10:03dionoeayeah, it's equivalent except that setcookie allows setting the experiation (while the array method doesn't)
22:10:07kugelthen just if(isset($_COOKIE["prevent_abuse"])) "Stop abusing";
22:10:38 Join webguest [0] (
22:10:41bluebrotherdomonoky: what do you think about completely dropping the turkish and chinese translations? They are in really bad shape and turkish itself is broken on file level
22:10:54dionoeawe basically do $_COOKIE["skinrated_{$_POST["skin_id"]}"] != $_POST["skin_id"] that way people can only rate a skin once
22:11:02bluebrotheror should we rather remove them from the resource file to exclude them from the binaries?
22:11:08 Join froggyman [0] (
22:11:24domonokybluebrother: removing them from the resource is good.
22:11:43kugeldionoea: I think we want download count to count again after some time
22:11:43 Quit webguest (Client Quit)
22:11:54dionoeacookies timeout
22:12:01domonokyand maybe remove the broken turkish, but chinese should stay in svn to ease a future update.
22:12:33domonokykugel: use two cookies, one with short timeout for download counting, and one with long/ever for rating.
22:12:38dionoeakugel: it looks like the default timeout is a browser session
22:12:48dionoeawhich seems decent enought
22:13:04 Quit phanboy4 (Ping timeout: 248 seconds)
22:13:24rasherbluebrother: Why would you drop a translation simply for being in bad shape?
22:13:26JdGordon_the rating cookie only lasts untill the update though yeah?
22:14:35*kugel has a look
22:14:42domonokyJdGordon_: if we use the themeid as cookie value, its not a problem. updates get new ids.
22:15:13JdGordon_just makign sure :)
22:15:26 Join FOAD_ [0] (
22:15:30bluebrotherrasher: well, is a translation that has less than 40% of all strings translated actually useful? At least I prefer untranslated applications rather than reading two languages all the time
22:15:59rasherbluebrother: But you understand English perfectly. For someone who doesn't, 40% is pretty good
22:16:13bluebrotherand turkish itself is broken on file level −− all non-ascii characters are encoded as the same character. Which makes it really useless.
22:16:33rasherWell yes, that's no good obviously
22:16:52bluebrotherrasher: I wouldn't call my english anything about perfect. But admittedly, I'm not uncomfortable using that language :)
22:18:33kugelhrm, I don't get a cookie
22:18:57 Join yosafbri` [0] (
22:19:06 Join phanboy4 [0] (
22:19:13 Quit FOAD (Ping timeout: 272 seconds)
22:19:13 Nick FOAD_ is now known as FOAD (
22:19:13 Nick pamaury_ is now known as pamaury (
22:19:17kugelah wait there it is
22:19:51 Quit pyro_maniac_ (Remote host closed the connection)
22:20:04LloreanMaybe it would make sense to define some strings as "core strings" (being the minimum necessary to use the player; which would just be very basic things like the label for the file browser and some error messages) and as long as those are still translated leave "old" translations in?
22:20:06kugelthe expiration time doesn't seem to work
22:20:57kugelor firefox doesn't show it properly...
22:22:57kugelok, that was really easy
22:23:44Mode"#rockbox -o Llorean" by ChanServ (ChanServ@services.)
22:24:48kugeldomonoky: that blocks downloadcount += 1 for 30s
22:25:04CIA-88New commit by 03bluebrother (r24604): Rockbox Utility russian translation update. ...
22:26:26domonokykugel: shouldnt it check for the theme id ? so it counts if i download different themes ?
22:26:41kugelah you mean it should be on a per-theme basis?
22:26:49dionoeathat would be coherent
22:27:05kugelmakes sense
22:28:05domonokyalso i would name the cookie a bit different (downloadcnt or similar) so we dont get conflicts when we add a cookie for the ratings.
22:29:16kugelyes, I didn't plan to leave it at "abuse" actually :)
22:31:47kugelhm, that means mutiuple cookies, else downloading two themes in an alternating fashion still permits abuse
22:32:44domonokyjust sprintf the themeid into the cookie name.
22:32:55kugeljup did so and it works
22:34:17*kugel needs to push his themes before this goes in!
22:34:54domonokyif it works, but it in.
22:35:39domonokyone small "bug" downloadcnd :-)
22:35:55kugelyep, saw it :) is 30 ok or do we want a higher value?=
22:36:28domonokyit could easily be more and wont hurt.
22:36:41*JdGordon_ whips up a script to download my theme every 31 seconds
22:37:18bluebrotherwhat are you currently implementing? Hammering prevention?
22:37:29 Quit Omlet (Read error: Connection reset by peer)
22:37:37domonokybluebrother: abusing download counts.
22:38:03domonokyyou can still download the theme multiple times in 30 seconds, it just wont be counted.
22:38:34bluebrotherah, ok. Just was wondering if it would conflict with rbutil's ability to download multiple themes :)
22:39:02 Join Range_304 [0] (
22:39:16 Part Range_304
22:39:24kugeldo rbutil downloads count at all?
22:39:46*kugel goes for 3min
22:39:48domonokyyes, they count too.. i also give rbutil the download.php link
22:40:35bluebrotherreminds me that I wanted to display the theme name instead of the download filename in the progress dialog. With that script link it became pretty ugly (and longish)
22:40:42domonokysee rbutil.tpl
22:41:14pixelmaJdGordon_: on my M5 I get something on the default radio screen without an fms set - but it looks a bit messy. First line reads literally:
22:41:17pixelmaStation: %d, %02d MHz (and then it's showing the current frequeny)
22:41:42JdGordon_you didnt update english.lang then?
22:41:57JdGordon_and yes, thats the other minor fix thats still needed
22:42:12pixelmaI did a complete build and unzipped it fully
22:42:12JdGordon_I need to deprecate LANG_FM_STATION and come up with a new one for "Station:"
22:42:41JdGordon_ok thats odd, english or another language?
22:42:45pixelmaoh, hmm. That's with the previous patch
22:43:01pixelmacurrently German
22:43:16pixelmaand so it's showing "Sender:" actually
22:43:27JdGordon_that could be the problem then
22:43:27CIA-88New commit by 03kugel (r24605): Prevent abusing download count a bit using cookies, downloads don't increment the counter until 3min have passed after the previous counted one
22:43:45pixelmaI don't like the bare look of the default radio screen though
22:43:45JdGordon_kugel: 3min? thats a bit drastic
22:43:56JdGordon_pixelma: isnt t the same as svn?
22:44:04kugelis it?
22:44:10pixelmaerr... no
22:44:26 Quit petur (Remote host closed the connection)
22:44:52domonokyJdGordon_: why is that drastic ? its only the counter and only if you download the same theme again and again.
22:45:03 Join petur [0] (~petur@rockbox/developer/petur)
22:45:17pixelmathere is no progressbar as far as I remember but more info about mono/stereo and preset/scan mode - need to recheck but it's definitely more than 3 lines
22:45:30JdGordon_domonoky: oh, you can keep downloading, just the coutner doesnt inc?
22:45:40domonokyJdGordon_: sure
22:45:45JdGordon_ok then its fine
22:46:10 Join flydutch [0] (
22:46:23domonokynow we need todo the same for rating, just with a verylong timeout. :-)
22:46:43*JdGordon_ thinks this is all wasted effort
22:46:53B4gderjust note that doing the same operation with curl and wget won't get stopped by that cookie thing
22:46:54JdGordon_can you make it show a listing of what is included in each theme?
22:47:12pixelmaJdGordon_: ok, the first line problem really has to do with the language but that shouldn't happen if the patch goes in...
22:47:32JdGordon_yes, see about about deprecating that string
22:48:10domonokyB4gder: sure, it isnt bullet proove, but probably good enough.
22:48:27kugelan updated theme is given a new id, right?
22:48:28B4gderyeah, it'll prevent the repeated pretting of the link a browser
22:48:39B4gderuh pressing
22:48:43domonokykugel: correct, a updated theme is technically a new theme.
22:49:15domonokyand theme ids are just the line numbers of the theme table.
22:49:54pixelmaSVN is four lines on swcodec: (1) Preset number. Preset name (2) Station: frequency (3) Mono/Stereo (4) Mode: scan/preset
22:50:13 Quit petur (Remote host closed the connection)
22:50:52 Join petur [0] (~petur@rockbox/developer/petur)
22:51:52pixelmahwcodec also has the peakmeter below
22:52:06domonokykugel: maybe we should reset download counters and ratings after we have implemented both abuse measures ?
22:52:23kugelI have thought about it too
22:52:30kugelwe could, but I don't care much
22:52:38***Saving seen data "./dancer.seen"
22:52:48*pixelma has 3 radios running currently ;)
22:53:26kugelawesome: "DB Error: not an error (0)"
22:53:47*kugel posts that to tdwtf
22:56:16kugelok, that works
22:57:09Unhelpfulamiconn: i had put *some* thought into making the audio buffer (including buffer_alloc stuff done before playback) buflib-like. the idea i had was to add a flags or types bitmask to buflib allocations, with certain flags indicating that certain types of data would exist before the actual buffer item. one of these flags would've marked "relatively static" data that was not supposed to be moved without calling a callback to let its owner know
22:57:09Unhelpful. the idea was that "static" allocations could replace most of buffer_alloc uses, but we would still be able to make changes to these, for example to get a larger buffer on skin load, etc.
22:57:20 Part Dhraakellian ("Meddle not in the affairs of ircops, for they are (not so) subtle, and quick to anger.")
22:57:22JdGordon_pixelma: preset name/number doont really mean anything in scan mode
22:57:45Unhelpfulalso a bunch of stuff the audio buffer has now, like filename, would be gone except for items that actually want it.
22:58:33 Quit evilnick_B (Quit: Page closed)
22:58:47pixelmaJdGordon_: yes, and?
22:59:06JdGordon_so having them on the screen is pointless?
23:01:18CIA-88New commit by 03kugel (r24606): Prevent abusing rating a bit by using cookies, this time we try to prevent rating a theme mulitple times entirely (which basically can't work with ...
23:02:34domonokykugel: nice commit message :-)
23:02:49pixelmacurrent SVN shows a blank line instead. That's not a reason to combine the mode line with it - especially as there are only 4 lines at most and I think it looks a bit better as it makes better use of the scree, especially if you think about the fact that preset number and name get more room (hence less scrolling) when not combined with the mode line
23:03:30 Quit bertrik (Quit: De groeten)
23:03:33pixelmaand preset mode is what I use much more than scan mode
23:05:16kugeldomonoky: so do we reset now?
23:05:29 Join kramer3d_ [0] (~kramer@unaffiliated/kramer3d)
23:05:46domonokykugel: just done it :-)
23:06:44gevaertsamiconn: I agree that in general plugin authors should make sure they don't leak, but right now we have lots of plugins that may or may not be buggy. Maybe we should just leave it enabled for a week or two, and then disable it again?
23:08:30 Quit kramer3d (Ping timeout: 256 seconds)
23:09:04 Quit dmb (Ping timeout: 256 seconds)
23:09:56 Join anewuser [0] (anewuser@unaffiliated/anewuser)
23:10:02 Join stripwax [0] (
23:10:05pixelmahmm... my .fms doesn't seem to be loaded. The cfg file should contain a line "fms: /.rockbox/WPS/my.fms" and my.fms should be at the place specified. correct?
23:10:59*petur has the recording histogram finally working in the sim \o/
23:11:16pixelmaprovided I use my.wps too (I actually copy-pasted the wps line in the config and edited the file extension and the beginning of the line
23:11:37pixelma^ JdGordon_
23:12:17JdGordon_lower case wps in the setting
23:12:33JdGordon_unless our file code is case insensitive
23:12:39JdGordon_that is correct otherwsie
23:12:40pixelmayes, caps lock error
23:13:09pixelmawhen typing it here I mean
23:13:15gevaertsuhm, it's case sensitive?
23:14:11pixelmadon't know and doesn't matter as it is all lower case on target
23:14:13*gevaerts thinks that if so, that's a bug
23:14:31gevaertsah, ok
23:14:32JdGordon_I dont think it is
23:14:47*gevaerts shuts up
23:14:53 Nick kramer3d_ is now known as kramer3d (~kramer@unaffiliated/kramer3d)
23:16:21CIA-88New commit by 03stripwax (r24607): Get rid of ffmpeg_scratchpad member (not needed since previous commit - accidentally left it in)
23:17:20kugelthe playlistviewer is annoying
23:17:31JdGordon_in what way?
23:17:41 Join mt [0] (~mtee@rockbox/developer/mt)
23:17:51kugelin the way that pressing down opens the 2nd context menu instead of going to the main menu
23:18:33pixelmawhich target?
23:18:58pixelmaand huh?
23:19:13kugelfuze, e200 and probably more
23:19:36CIA-88New commit by 03bluebrother (r24608): Don't close the disk handle to the ipod too early. ...
23:19:48pixelmacan't follow your description what it does or not
23:21:21pixelmawhich might be related to the fact that I don't have a scrollwheel target and I currently think that it is this way on those targets
23:21:52 Quit bluebrother (Ping timeout: 265 seconds)
23:22:02kugelpossible, but the playlist viewer has 2 context menus
23:22:15kugeland one of them overrides the global go-to-mainmenu button
23:22:51 Quit GeekShado_ (Quit: The cake is a lie !)
23:23:22 Join bluebrother [0] (
23:23:23 Quit bluebrother (Changing host)
23:23:23 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
23:23:24 Join Adnyxo [0] (
23:24:41bluebrotherAlexP: that was the bug discussed in the forums. I'm still considering not to answer in that thread.
23:25:01pixelmatrue. I never noticed as I never wanted to go to the main menu from there, just view or manipulate the playlist or get back to the WPS
23:25:22pamauryIs there any tagcache expert here ?
23:25:35 Part _zic1
23:25:36 Quit Horscht (Ping timeout: 264 seconds)
23:26:39CIA-88New commit by 03bluebrother (r24609): Remove turkish from the languages bundled in the binary. The translation is broken on file level (all non-ASCII characters are broken).
23:31:12 Quit fml (Quit: CGI:IRC)
23:32:31pamauryHow does rockbox detect the type of a music file ? Is it only extension based ?
23:32:39kugelpamaury: I'm no expert but I have looked at the code time ago
23:32:48pixelmaJdGordon_: in my current preset list, there are 6 of 17 stations that don't fit completely on the line with "Preset: " in front which fit before - and the line doesn't even scroll
23:33:13pixelmathat's on my M5 with 12-Nimbus
23:33:20domonokypamaury: it starts with the file extension. this maps to a metadataparser, which can change the codec type again.
23:33:26kugelpamaury: IIRC it's primary extension based, but the metadata parser can change that
23:33:41pamaurykugel: I'm not sure. Suppose I have a complete file path P and I want to retrieve some properties of the tagcache entry associated to this name. How can I do that ? Do I use the tagcache_find_index function ?
23:34:16pixelmaand it seems I can't load an own fms, or I did something wrong but you didn't tell me before
23:35:02kugelpamaury: get_metadata() IIRC
23:35:42pamaurykugel: yes but then if I want to be able to modify some data associated to the file, then it's better to use tagcache (it's for MTP), no ?
23:35:59*domonoky thinks get_metadata() wont use tagcache, but directly parse that file.
23:36:16pixelmapamaury: it doesn't find the mp3s put on my c200 in MTP mode with the SansaOF which are renamed to <some arbitrary string>.dat but work fine if I change the extension back to .mp3
23:36:23kugeltagcache doesn't allow modifying the files IIRC
23:36:51JdGordon_pixelma: ok, so the %s needs to be added
23:36:54pixelmaso I *assume* from a user perspective that it's extension based
23:36:55pamauryno, but it allows to modify some data associated to it like the rating no ?
23:37:14kugelrating and playcount, that's all
23:37:25kugeland you need gather runtime data enabled for that
23:37:41kugelthe other meta data is read-only afaik
23:38:06pixelmaJdGordon_: I would also prefer these things broken into two lines again
23:38:07pamauryOk, so assume that I code a read-only version of MTP, then you think get_metadata is better than tagcache to retrieve meta data ?
23:38:40JdGordon_pixelma: would you like to write the code and ill copy it into the .c?
23:38:58 Join dmb [0] (~Dmb@unaffiliated/dmb)
23:39:00kugelpamaury: if domonoky is right then get_metadata() maybe isn't the best candidate
23:39:22pamauryYes, because it will be awfully slow
23:39:30domonokypamaury: get_metadata() will work even without tagcache, but it will be slooow
23:39:50pixelmayou already have it almost, just shuffling it around a bit. And I'd like a way to test (other than recompile ;) )
23:40:13*pixelma could also try on c200, maybe things are also different there...
23:40:18pamauryThen, I can suppot both. But tagcache is mandatory on all targets, on ?
23:40:19CIA-88New commit by 03b0hoon (r24610): Packard Bell Vibe 500: Start committing plugin keymaps.
23:40:36kugelpamaury: if ramcache is available then I'd use tagcache_fill_tags()
23:40:56pamauryHum, interesting
23:41:13pixelmaJdGordon_: I could do that of course. Btw. the progressbar is quite nice
23:41:16kugelyou need to perform a tagcache search for it though
23:41:47pamauryIt a tagcache search slow ?
23:41:57JdGordon_it is if its not in ram
23:42:14kugelpamaury: err no, you dont
23:42:18kugelbool tagcache_fill_tags(struct mp3entry *id3, const char *filename)
23:42:20domonokypamaury: tagcache is not avaiable be default. users need to initialise the database first.
23:42:59kugelI'm not sure whether tagcache is useful for mtp
23:43:12pamauryThat's my current question :)
23:43:56kugelI rather doubt it
23:44:26kugelmtp would be pretty useless if you could really only transfer audio files with it
23:44:30pixelmatagcache still chokes on large collections, right? Also with no music files (can't remember seeing a fix at least=
23:45:05kugelpixelma: someone noticed strange behavor with 65k+ songs, and I fixed that
23:45:25kugelthe upper limit is theoretically 4 billion now
23:45:41domonokyand would .databaseignore files also be effective for mtp ?
23:45:50pamauryThe possible use I see of tagcache is to retrieve metadata. Indeed, MTP provides a way to retrieve metadata without retrieving the whole file. Here tagcache helps because it contains all the necessary data.
23:46:16pixelmahmm... maybe I confused it with dircache
23:46:19pamaurydomonoky: what is .databaseignore ?
23:46:36pamauryI already use dircache :)
23:46:38kugelpamaury: in that case, it may be useful
23:46:44domonokya file to let the database ignore files in certain directorys.
23:46:50pixelmait's database.ignore
23:47:25domonokythere is also database.unignore :-)
23:47:39amiconnJdGordon_, gevaerts: vfat is case preserving, but case insensitive
23:47:40pamauryok, I didn't know it existed :)
23:47:56pixelmait's more of a :( though
23:48:19 Join Casainho [0] (
23:48:20pamauryWhy ?
23:49:05 Quit Adnyxo (Quit: Leaving)
23:49:53domonokyif other mtp features still work for everthing, it might be ok to just provide metadata access for those which are in the db.
23:50:08pixelmabecause it slows things down again. Before the unignore tagcache didn't look deeper into the directory structure if you had a .ignore at the top level. Now it still goes ahead looking into subfolders searching for an unifnore
23:50:26domonokyand mtp could ofcourse update tagcache while transfering new files :-)
23:50:37 Quit flydutch (Quit: /* empty */)
23:52:40 Quit jgarvey (Quit: Leaving)
23:52:43pamauryOk, I think my code will try to use tagcache_fill_tags if the db is in RAM, then tagcache_find_index if not and finally get_metadata if db is not initialized. I think performance will get worse in this order
23:53:51 Quit efyx_ (Remote host closed the connection)
23:53:57 Join checker [0] (
23:54:50checkeranyone know what "BM6" means? My wps isn't showing up, instead i get this little code
23:55:34JdGordon_pixelma: ok, yeah, if you make the default fms I'll happily use it. I'll work on trying to figure out the last issues tonight
23:55:34 Quit ender` (Quit: The problem with political jokes is they get elected. -- Henry Cate, VII)
23:56:59pixelmaand I would be opposed to this going in without a fms browser, editing the cfg is a bah
23:57:34JdGordon_yes yes, dont worry about that
23:57:50JdGordon_its not done yet because its the boring finiky bits at the end which I just havnt done yet

Previous day | Next day